Eco - friendly MoS 2 /waste coconut oil nano ﬂ uid for machining of magnesium implants

,


Introduction
Recently, alloys have been employed in manufacturing medical implants, particularly orthopaedic implants. In orthopaedic implants, biodegradable magnesium alloys are widely and usually preferred over stainless steel, titanium alloys, and cobalt-based alloys for avoiding undesirable outcomes like metal ion release and stress shielding as well as reducing the cost and weight [1]. For critical bone fracture, permanent alloy implants are needed and usually titanium alloys and steels are preferred, though they are expensive and heavy [2,3]. As Mg alloy is the lightest metal (1.738 g·cm −3 density [4]) and compatible with the structural application for bones (as critical bone density is in the range of 1.75-2.10 1.738 g·cm −3 [1]), in last decade the research on Mg alloys augmented 491% [5]. Pure Mg is less corrosive (407 mm·year −1 ) than its alloy form [6]. The Mg alloy with aluminium and zincreinforced composite was synthesized and tested for Table 1. The machinability could be improved by improving the tool hardness either by using coated tool [27,28] or by changing a much harder tool for machining. Alternatively, it can be done by reducing the tool wear. The tool wear could be reduced by supplying coolant and lubrication appropriately at the cutting zone. Coolant cost is one of the considerable costs in manufacturing. The literature suggested eco-friendly flood cooling [29], ultrasonically atomized cutting fluid [30], minimum quantity lubrication (MQL) [31,32], nanoparticles mixed flood cooling [33], hybrid nanoparticles mixed flood cooling, nanoparticles mixed MQL [34][35][36][37][38][39], hybrid nanoparticle mixed MQL [33,[40][41][42], cryogenic cooling [43][44][45], and cryogenic/lubrication hybrid cooling [46]. The research gap can be stated as follows: all proposed methods in the above-discussed literature would lead to additional machining costs. The MQL is costlier than flood cooling and cannot be reused. The addition of nanoparticles in the MQL mode leads to the additional cost of the coolant. Nanoparticles may mix in the environmental air after drying the coolant if not cleaned properly.
Hence, this investigation was innovatively approached to develop a cheap and effective coolant from the waste resource. A reusable nanofluid developed from the waste coconut oil was assorted with MoS 2 nanoparticles to improve the machined surface quality, and reduce the tool wear, the cutting force, the feed force, and the cutting zone temperature. As this coolant contains edible oil, it is biodegradable and utilizes waste disposal. The performance of the proposed nanofluid is compared with the conventional coolant and it is found to be costlier. The commercial coolant was mixed with water. This coolant is 100% water-free and utilizes the used-edible oil of coconut oil (waste) for a valuable machining process.   [31] Poor machinability in C NC turning of AISI H13 tool steel TiN-coated tool Uncoated carbide tool for life tool wear Less tool wear compared to the conventional uncoated carbide tool [27] Poor machinability of C NC turning of EN31 alloy steel

TiN-coated tool
Uncoated H SS tool Higher chip thickness [28] Tool wear in machining titanium alloy Ti-6Al-4V

MQL for machining
Dry and flood coolant The edge chipping and adhesion of chips to the cutting tools were reduced [32] Improper cooling at the cutting zone while machining alloys Nano-additives: MQL for machining Compared with MQL Nano-additive-MQL system was found to be sustainable machining [47] To optimize the use of coolants instead of conventional flood cooling MQL is composed of biodegradable oil with various nanoparticles for machining Compared with dry and MQL Achieved better surface finish, low power consumption, and fewer tool wear [48] To achieve environmentally sustainable manufacturing in pressure vessel carbon steel (SA516) using coated carbide inserts MQL for machining Compared with traditional flood coolant Surface roughness and tool wear reduced significantly [49] Drilling on titanium Cryogenic cooling conditions Compared with traditional flood coolant Drill maximum temperature reduced by 33-50% [43] Progressive tool wear while machining titanium alloy Ti6Al4V due to built-up edges, diffusion wear, and cutting forces Cryogenic cutting with liquid nitrogen Compared with conventional dry cutting and flood cutting Eliminated the issues of BUE formation, diffusion wear, and cutting force fluctuation, especially under aggressive cutting conditions than flood cooling [44] Poor machinability for Ti-6Al-4V thin-wall components Cryogenic MQL MQL Highly improved machinability, superior lubrication, and high cooling function [45] To develop sustainable and optimized machining of Ni-based industrial alloy Adequate cooling/lubrication Flood, MQL, and cryogenic Improved surface quality and support for sustainable means of production [46] Machining of Hastelloy C-276

MQL for machining
Compared with conventional dry cutting and flood cutting Cooling environment influencing cutting zone temperature, surface finish, and material removal rate [50] Oil mist generation and endanger the health of workers This research aims to develop a low-cost, high-performance nanofluid for machining biodegradable, biocompatible, and safe manufacturing of circular Mg implants with the use of eco-friendly nanofluid coolants. This research discusses improving the machinability performance and optimizing the process parameters for biocompatible magnesium implant manufacturing for biomedical applications with eco-friendly nanofluid of MoS 2 nanoparticles suspended in waste coconut oil. The machinability performances are observed under flood cooling conditions and compared with the conventional commercial coolant.
Cutting fluid costs and the risk of disposing of them are major issues in flood cooling systems in metal machining processes. This investigation overcomes those issues by proposing to develop a coolant from the waste resource. Reusable nanofluid was developed from waste coconut oil and assorted with MoS 2 nanoparticles to improve the machined surface quality and reduce the tool wear, cutting force, feed force, and cutting zone temperature. As this coolant contains content edible oil, it is biodegradable. The performance of the proposed nanofluid is compared with conventional, costlier, and commercial coolant oil mixed with water. This coolant is 100% water-free and utilizes multiple times used edible coconut oil (waste) for a valuable machining process. Hence, the novelty statement is to develop an effective nanofluid for flood cooling of machining Mg implants; it was prepared by mixing waste oil (used coconut oil) and MoS 2 nanoparticles at a low concentration, experimenting with a real-time application, statistically evaluating observations and comparing them with conventional coolant's performance, and then optimizing the process parameters for computerised numerical control (CNC) machining of Mg implants with the help of Taguchi analysis and analysis of variance (ANOVA). The research background, the literature review, the research gap, and the novelty of this investigation are presented in Section 1; the materials and research methods followed and observations are illustrated in Section 2. The results are analysed and optimized for minimizing cutting force, feed force, surface roughness, cutting zone temperature, and tool wear and explained in Section 4. The findings and conclusion are based on the results of the experimental observational analysis and are presented in Section 5.

Materials and methods
The flow of this investigation is presented in Figure 2. This investigation was carried out at the in-house research facility of high precision heavy duty (5HP) lathe ( Figure 3).
The facility offers a high degree of accuracy in machining and can operate at eight different speeds in the range of 32-1,200 rpm and 18 variety automated feed (0.025-2.5 mm·rev −1 ) conditions. Medically pure Mg rods were utilized as samples for testing the machinability. The round rods of the Mg workpiece material were purchased from Pujara High-Quality Steels Pvt. Ltd, Chennai. The mechanical properties of Mg implant materials, namely the density and elastic modulus of magnesium implants were 1.82 g·cm −3 and 43.5 GPa, respectively, and it is equivalent to the mechanical properties of bone [12,53,54]. The sample length of machining was 50 mm long. The conventional cutting fluid is Castrol make, Syntilo 9930 grade Downers Grove, Synthetic Coolant type IL mixed with 5% water. For preparing the nanofluid coolant, the MoS 2 nanoparticles were obtained from the chemical laboratory of Saveetha School of Engineering and reduced in particle size by grinding in a ball mill for 15 h. The waste coconut oil was obtained from Kerala Chips at Vadapalani, Chennai Branch, and filtered. The quantity of MoS 2 added was limited to the viscosity requirements of the coolant oil so that the pump power would not affect and the rated flow could be achieved. The waste coconut oil (used coconut oil) was filtered and then the impurities were removed. By the iterative test, the 30 g per 1,000 g of waste coconut oil ratio was finalized. The inclusion of MoS 2 not only improved the cooling and lubrication properties but also improved the fire point of the nanofluid. The nanofluid was prepared by mixing oil and nanoparticles of MoS 2 (<5 nm in size) in the ball mill for 12 h, which allows the nanoparticles to float on the mixed oil, which is a nanofluid. The coolant flow rate was set at 3.5 L·min −1 and it was ensured that no fumes were generated while cutting and the proposed oil is biodegradable. As no human sample is involved in this research, no ethical approval was required for this investigation. Hence, MoS 2 nanoparticles enriched with a waste coconut oilbased nanofluid in the flood-cooled machining process were verified and found to be very safe. The properties of the prepared nanofluid coolant were characterized and are furnished in Table 2. The mechanism behind the machining is cut by shearing and exposes the inner layer to the atmosphere at elevated temperatures [55,56]. If the water content coolant reacts with the surface, oxides may be produced and fumes generated. Hence, for this nanofluid, lubrication properties were considered to be a primary concern for the coolant (like coconut oil and MoS 2 ) preparation.
The heat was measured as the temperature at the cutting zone. The mercury pool thermocouple technique is employed for the temperature measurement; mercury was filled in the blue colour container (refer to Figure 3, left).
The force encountered while machining was measured with a lathe tool dynamo meter, which operates with 12 strain gauges. Each strain-gauge possesses 350 Ω resistance. The gauge factor is 2 ± 1. The set-up measures forces in three directions and converts them into net force acting on the tool as output.
The surface roughness was measured by using SUR-FTEST SJ-410. Figure 4 shows the surface roughness tester (Mitutoyo, Japan), which is a portable type surface roughness tester, employed to measure the surface roughness with a setting of 4 mm inspecting length, 0.5 mm·s −1 speed, 0.8 mm cut-off length, 2 mm nose radius, 600°angle of the tip, and 6 and 3.5 mm height and width of the stylus, respectively. The precision balance of LC 0.01 mg was utilized to measure the tool wear (in mg·h −1 ) by dividing the machining time from the mass loss by the tool.

Experimentation
The limiting values of input variables are fixed with trails at various feeds, speeds, and depths of cut employed while finishing the shaft manufacturing in industries at Ambattur, Chennai. The variable input parameters are cutting velocity (in m·min −1 ), nose radius (in mm), and feed rate (in mm·rev −1 ). The range and levels of parameter variation are shown in Table 2 [57][58][59][60][61]. The carbide tip tool was employed in machining all samples in all kinds of cutting environments and the fresh tool was used in each experiment. The L16 type experimental design was used for the choice of the Taguchi experimental design for three factors at four levels. The experiments were conducted accordingly, and observations of the cutting force, feed force, surface roughness, cutting zone temperature, and tool wear were noted. All measurements were taken at five different locations/stages of processing of the workpiece after/during experiments, and the average was taken at each experiment [62,63]. The observations are consolidated for the control group in Table 3 and those for the intervention group in Table 4.

Results and discussions
The properties of the prepared nanofluid coolant were characterized and are given in Table 5. The machinability improvement was ensured by analysing the results with the help of independent sample tests and Taguchi and ANOVA procedures with the responses of cutting force, feed force, cutting zone temperature, surface roughness, and tool wear. The inputs are common except for the coolant of wet machining for the control group and intervention group.

t-Test
The machining performance was investigated for both machining methods by comparing the responses like cutting force, feed force, cutting zone temperature, surface roughness on machined surfaces, and wear of the tool. These results were statistically compared by performing one-way ANOVA, and a statistically significant difference  was observed for material removal rate (p = 0.026, p < 0.05). Table 4 shows group statistics that are the results of the t-test. From this table, it is clear that the average (mean) cutting force reduced from 280.0863 to 211.8546 N with the use of the proposed nanofluid in place of a conventional commercial coolant ( Figure 5). Similarly, the mean feed force decreased from 124.5194 to 90.3394 N ( Figure 6), the mean cutting zone temperature from 263.5150 to 203.0800°C (Figure 7), the mean surface roughness from 0.36278 to 0.24387 µm (Figure 8), and the mean tool wear from 0.7525 to 0.3531 mg·h −1 (Figure 9). The above-discussed values are their mean values; Table 6 shows the improvement in the machinability average by using a nanofluid-based coolant obtained from the machining practice than the dry machining practice. Table 7 shows the results of the independent sample tests. It is evident from the table that the observations are significant, i.e. they did not violate statistical  assumptions. As the value of significance (p) is 0.001 and it is less than 0.05, for cutting force observations of both groups (control and intervention groups), it can be confirmed that the observations obey statistical assumptions and are acceptable. Similarly, the significance value (p) is 0.023 (p < 0.05) for feed force observations. It was observed that p = 0.0001 and p < 0.05 for the cutting zone temperature; p = 0.045 (p < 0.05) for surface roughness; and p = 0.028 (p < 0.05) for tool wear. Hence, observations were significant and acceptable as per the results of the independent sample tests.
As shown in Table 7, the positive mean values comparatively decrease the respective values than in the control group. It is understood that the cutting force at t 4.682 = 68.23167 N, i.e. 68.23167 N cutting force, averagely reduced with the use of the proposed nanofluid in place of a conventional commercial coolant; similarly, the feed     force at t 22.933 = 34.18000 N, cutting zone temperature at t 3.758 = 60.43500°C, surface roughness at t 13.319 = 0.118908 µm, and tool wear t 4.660 = 0.39938 mg·h −1 . Hence, the proposed method of processing significantly improved machinability. Figure 5 shows the G-graph output by comparing the mean values of the cutting force encountered at a 95% confidence level and error ±1% in the control group and the intervention group. Similarly, Figures 6-9 show the G-graph output by comparing the mean values of the feed force, cutting zone temperature, surface roughness, and tool wear.
The minimum value of the responses, such as responses of the cutting force, feed force, cutting zone temperature, tool wear, and surface roughness, are preferred in the machining metals in the CNC lathe; therefore, the minimum value is best for obtaining the high-quality product, safe and reliable performance for a long time too [64][65][66]. Here, the optimization is to minimize the responses of the cutting force, feed force, cutting zone temperature, tool wear, and surface roughness, and the above-said responses are considered. In this investigation, it was suggested to minimize the responses; Hussain [53] and Naqiuddin et al. [54] proposed using the following signal-to-noise (S/N) ratio equation, where smaller values are preferred: where the number of trails is expressed as N j , and the subscripts j and v denote the trail and test values.

Taguchi analysis of the cutting force
Taguchi analysis gives the degree of possibilities to achieve the objective function. Here, the objective is to minimize the cutting force (the smaller the better) [21][22][23][24][25][26][67][68][69][70]. For this statistical model, the intervention group observations of cutting forces on 16 different experiments were used to optimize the proposed method. Figure  10 shows the main effect plots for the signal-to-noise ratio for the response observations of the cutting force during experiments with the proposed nanofluid. The above mean line indicates large signals (possibilities towards the objective function of reducing the cutting force) and the left-most graph for the cutting velocity shows an increase of the cutting force with the increase of the cutting velocity; in other words, a decrease of possibilities (signals) in minimizing the cutting force [71][72][73]. The decision could be made based on the highest signal-to-noise ratio. Hence, the input setting of the cutting velocity of 20 m·min −1 (S/N = −43.23), the feed    Figure 10). The variability is found to be higher in the cutting velocity input than in the tool feed and significantly less in the nose radius of the tool used [74][75][76]. Table 8 shows the results of the Taguchi analysis, which is the yield of S/N ratios. Table 8 shows the classification of the process parameters based on their influence according to the delta value obtained in this statistical analysis. The higher delta value indicates a higher influence based on the delta value ranking of the process variable determined. As the delta value for the cutting velocity is 5.58 and greater than the data value of the tool feed and nose radius factors, the cutting velocity is the no. 1 factor (Rank 1) and influences the cutting force while machining. Accordingly, Rank 2 is for the tool feed (delta = 1.41) and Rank 3 is for the nose radius of the tool.

ANOVA results on the response of the cutting force
The residual plots for cutting force observations are shown in Figure 11. The normal probability plot shows the quality of observations [77][78][79][80][81][82]. No observations deviate much from the mean line and residual error in these observations. Hence, it is concluded that the observations are not violating the statistical assumption and are ensured as valid. Table 9 shows the results of the ANOVA on the response of the cutting force for the decision on the influence of factors. In this stage, one can decide the factor level based on the p-value. Here, a minimum p-value indicates more influence. The decision criteria are as follows: if p < 0.05, the factor is significantly influencing the measured response, and p > 0.1 indicates an insignificant factor [83][84][85]. From the Taguchi analysis, it was observed that the cutting force was of Rank 1 and the significant value was 0.001 (very low); similarly, p = 0.004 (p < 0.05) for the tool feed. But, in the case of the nose radius, p = 0.940 (p > 0.1), so it was found to be insignificant.   Table 9 shows the results of ANOVA at factor-level decision-making. This is a process that can be controlled for meeting the cutting force by altering the independent variables in which a highly sensitive independent variable has a low value of p; in other words, the independent variable which contributes more to alter the response of the cutting force has a higher F value [86][87][88]. Hence, the high contribution (F-240.72), a highly sensitive independent variable, is the cutting velocity (as F = 240.72 and p = 0.001 in Table 9). The next is the tool feed. But the nose radius is not significant as p = 0.940 > 0.10 [89][90][91], which means the change in the nose radius does not affect the cutting force considerably [92][93][94]. Hence, the contribution is very low (F = 0.13). The table of coefficients is shown in Table 10, from which one can take a deep decision on the degree of influence of factor at its level of input. Similar to Table 9, here the p-value indicates the significance of the level of factors on the response. From this output, the influence factors can be observed. As p > 0.1, the tool feed of 0.050 mm·rev −1 (p = 0.131) and nose radii of 0.3 mm (p = 0.832), 0.6 mm (p = 0.979), and 0.9 mm (p = 0.577) are not significant, as the nose radius does not influence. The relation between the tool feed rate and cutting speed is depicted on the threeaxis graphs surface plot in Figure 12. From this plot, the combined effects of both variables on the response of the cutting force at the cutting zone can be understood.

Regression equation
The regression equation was developed based on the coefficients obtained in the ANOVA, and the results are  shown in Table 11. The mathematical model is a mathematical equation from which anyone can interpret values of untested combinations or predict the process parameter levels for the desired response [95][96][97]. The developed regression model is given as follows: The 3D surface plot ( Figure 11) exhibits the relationship between the cutting velocity and tool feed inputs on the cutting force. The factor cannot be taken into account here as it is insignificant for the cutting force.
ANOVA results for validation of a statistical model for the cutting force response can be noticed from the R 2 value from the model summary ( Table 8). The condition is that the R 2 value must be greater than 95% for a good model; in this case, the value of R 2 is 99.22% and hence the model is acceptable. It also indicates that there is a good agreement between the predicted and experimented values.

Taguchi analysis of the feed force
As known, Taguchi analysis offers some degree of possibility to achieve the objective of the function. Here, the aim is to minimize the feed force (smaller the best). For this statistical model, the intervention group observations of cutting forces on 16 different experiments were used to optimize the process parameters. Figure 13 shows the main effect plots for the signalto-noise ratio for the response observations of the feed force during experiments with the proposed nanofluid. Figure 13 shows that a higher S/N ratio implies the best possibility for obtaining a low feed force at a cutting velocity of 20 m·min −1 , a feed rate of the tool of 0.025 mm·rev −1 , and a nose radius tool of 0.9 or 1.2 mm. The ambiguity of the nose radius tool selection is clarified in Table 12. Level 4 has the highest signal-to-noise ratio of −39.10 in comparison with both Levels 1 and 3, respectively.
So, a 1.2 mm radius offers a minimum feed force at the cutting zone.   The mean line in Figure 13 indicates large signals (possibilities towards the objective function of reducing the feed force); the left-most graph for the cutting velocity shows an increase of feed force with the increase of cutting velocity; in other words, a decrease of possibilities (signals) in minimizing the feed force [98][99][100]. Hence, the input setting of 20 m·min −1 cutting velocity, 0.025 mm·rev −1 feed rate, and 0.3 mm nose radius tool gave a minimum feed force. The variability is found to be high in the cutting velocity input than in the tool feed and very low in the nose radius of the tool used [101][102][103]. Table 12 shows the results of the Taguchi analysis, which is the yield of S/N ratios. Table 12 describes the process parameters based on their influence according to the delta value obtained statistically. The higher delta value indicates a higher influence based on the delta value ranking of the process variable determined. As the delta value for the cutting velocity is 0.6, which is greater than the data values of tool feed and nose radius factors, the cutting velocity is the no. 1 factor (Rank 1) influencing the feed force while machining. Accordingly, Rank 2 is for the tool feed (delta = 0.27), and Rank 3 is for the nose radius of the tool (delta = 0.02).

ANOVA results on the response of the feed force
The residual plots for feed force observations are shown in Figure 14. The normal probability plot shows the quality of observations. No observations deviate much from the mean line, and residual errors in these observations are in an acceptable range. But only one observation out of 16 was found to deviate (Table 13). Hence, observational accuracy was ensured from the statistical output of ANOVA, as shown in Figure 14.
The results of the ANOVA on the response of the feed force for a decision on the influence of factors are explained in Table 14. In this stage, one decides factorlevel decisions based on the p-value. Here, a minimum p-value indicates more influence. The decision criteria are as follows: if p < 0.05, the factor is significantly influencing  the measured response; p > 0.1 indicates an insignificant factor. From Taguchi analysis, it was observed that the feed force was of Rank 1 and the significant value was 0.001 (very low); similarly, p = 0.001 (p < 0.05) for the tool feed. But, in the case of nose radius, p = 0.846 (p > 0.1) so it was found to be insignificant. Table 14 shows the results of ANOVA at factor level decision making. It can be understood from Table 14 that for controlling the process for obtaining the desired response of feed force, alter the cutting velocity than the tool feed rate. There is no use in varying the tool inserts with different nose radii [104][105][106]. The table of coefficients is shown in Table 15, from which one can take a deep decision on the degree of the influence of the factor at its level of input. Similar to Table 14, here, the p-value indicates the significance of the level of factors on the response. Table 15 gives much insight into the above decision that variation of tool feeds up to 0.025 is better than the range 0.025-0.050 for reduction of the feed force on the tool. Figure 15 shows the 3D surface plot of the relationship between the tool feed and cutting velocity in the feed force response.    Table 15 mainly helps to form the mathematical model by generating the regression equation. The mathematical model helps one to interpret values of untested combinations or predict the process parameter levels for the desired response. The regression model is given as follows:

Regression equation
The validation of a statistical model for the feed force response can be observed from the R 2 value of the model summary ( Table 16). The condition is that the R 2 value must be greater than 95% for a good model; in this case, the R 2 value is 98.94% and hence the model is acceptable. As R 2 > 95%, the regression equation is reliable and its prediction accuracy is high.

Taguchi analysis on the surface roughness
Taguchi analysis gives the degree of possibilities to achieve the objective function. Here, the objective is to minimize the surface roughness (the smaller the better). For this statistical model, the intervention group observations of the surface roughness in 16 different experiments were used to optimize the proposed method. Figure 16 shows the main effect plots for the signal-tonoise ratio for the response observations of the surface roughness during experiments with the proposed nanofluid. The mean line indicates large signals (possibilities towards the objective function of reducing the surface roughness), and the right-most graph for the cutting velocity shows that with an increase of surface roughness, there is a decrease in the cutting velocity, i.e. a decrease in the cutting speed possibilities (signals) increases the surface roughness [81,[107][108][109][110][111]. Hence, the input setting of Level 4,80 m·min −1 cutting velocity (as maximum S/N ratio is −17.37), 0.100 mm·rev −1 (Level 4) feed rate (as maximum S/N ratio is −18.05), and 0.3 mm (Level 1) nose radius tool (as maximum S/N ratio is −18.25) gave a minimum surface roughness ( Figure 16 and Table 17). The variability is found to be high in the cutting velocity input than in the tool feed and very low in the nose radius of the tool used [112][113][114]. Table 17 shows the results of the Taguchi analysis, which is the yield of S/N ratios. Table 17 ranks the process parameters based on their influence according to the delta value obtained in this statistical analysis. A higher delta value indicates a higher influence based on the delta value ranking of the process variable determined. Since the delta value for the cutting velocity is 1.71, which is greater than the data value of the tool feed and nose radius factors, the cutting velocity has the highest influence on the surface roughness during machining, ranking first (Rank 1). Accordingly, Rank 2 is for the tool feed (delta = 0.61), and Rank 3 is for the nose radius of the tool (delta = 0.15).

ANOVA results on the response of surface roughness
The residual plots for surface roughness observations are shown in Figure 17. The normal probability plot shows the quality of observations. No observations deviate much from the mean line, and residual errors in these observations are in the acceptable range. Hence, from Figure 17 and Table 18, it is evident that, except fifth observation, the remaining observations are acceptable as per the statistical verification by ANOVA.    Table 19 shows the results of the ANOVA on the response of surface roughness for the decision on the influence of factors. In this stage, one can take the decision on the factor level, based on the p-value. Here, a minimum p-value indicates more influence. The decision criteria are as follows: if p < 0.05, the factor is significantly influencing the measured response; p > 0.1 indicates an insignificant factor. From Taguchi analysis, it was observed that the surface roughness was of Rank 1 and the significant value was 0.001 (very low); similarly, p = 0.001 (p < 0.05) for the tool feed. But, in the case of the nose radius, p = 0.090 (0.05 < p < 0.1), and it did not influence considerably. Table 19 shows the results of ANOVA at a factor level decision making, which reveals that for obtaining the desired surface roughness, it is better to alter the cutting speed for the best response and alter the tool feed for fine tuning. The alternation tools with various nose radii should be avoided.
The table of coefficients is shown in Table 20, from which one can take a deep decision on the degree of influence of a factor at its level of input. Similar to Table 19, here, the p-value indicates the significance of the level of factors on the response. Table 20 recommends the use of a cutting tool with a nose radius of 0.3 mm for obtaining a significant response on the surface finish (lower surface roughness). The recommended range of cutting velocity is 20-60 m·min −1 . Similarly, the tool feed from 0.10 to 0.025 mm·rev −1 will give appreciable results on the surface finish. Hence, the major contribution is the cutting velocity, and the tool feed rate inputs for the surface roughness and its interactive effects on the surface roughness on the job are depicted in Figure 18.   The validation of a statistical model for the surface roughness response can be observed from the R 2 value of the model summary ( Table 21). The condition is that the R 2 value must be greater than 95% for a good model; in this case, the R 2 value is 99.73%. Hence, the model is acceptable. The R 2 value also confirms the reliability of the mathematical model (Eq. (3)) in the prediction of the response of surface roughness.

Taguchi analysis on the cutting zone temperature
Taguchi analysis gives the degree of possibilities to achieve the objective function. Here, the objective is to minimize the cutting zone temperature (smaller is better in the signal-to-noise ratio). For this statistical model, the intervention group observations of cutting zone temperatures on 16 different experiments were used to optimize the proposed method. Figure 19 shows the main effect plots for signal-to-noise ratios for the response    observations of the cutting zone temperature during experiments with the proposed nanofluid. The mean line indicates large signals (possibilities towards the objective function of reducing the cutting zone temperature). The left-most graph for the cutting velocity shows an increase in the cutting zone temperature with an increase in the cutting velocity; in other words, a decrease of possibilities (signals) in minimizing the cutting zone temperature. Hence, the following input levels: level 1, 20 m·min −1 cutting velocity (as the maximum S/N ratio is −45.23); level 1, 0.025 mm·rev −1 feed rate (as maximum S/N ratio is −46.28); and level 4 of 0.3 mm nose radius tool (as maximum S/N ratio is −46.04) recorded a minimum cutting zone temperature (refer to Figure 19 and Table 22). The variability is found to be high in the cutting velocity input than the tool feed and very low in the nose radius of the tool used. Table 19 shows the results of the Taguchi analysis, which is the yield of S/N ratios.
The process parameters based on their influence according to the delta value obtained in this statistical analysis are shown in Table 22. The higher delta value indicates a higher influence based on the delta value ranking of the process variable. As the delta value for the cutting velocity is 5.58 and is greater than the delta values of the tool feed and nose radius factors, the cutting velocity is the no. 1 factor (Rank 1) and influences the cutting zone temperature while machining. Accordingly, Rank 2 is for the tool feed (delta = 1.41), and Rank 3 is for the nose radius of the tool.

ANOVA results on the response of the cutting zone temperature
The residual plots for cutting zone temperature observations are shown in Figure 20. The normal probability plot shows the quality of observations. The number of observations deviates from the mean line and the number of residual errors was observed in these observations. Hence, these observations are statistically acceptable.   Table 23 shows the results of the ANOVA on the response of the cutting zone temperature for a decision on the influence of factors. In this stage, one can take a decision on the factor level based on the p-value. Here, a minimum p-value indicates more influence. The decision criteria are as follows: if p < 0.05, the factor is significantly influencing the measured response, and p > 0.1 indicates an insignificant factor. From Taguchi analysis, it was observed that the cutting zone temperature was of Rank 1 and the significant value was 0.001 (very low); similarly, p = 0.001 (p < 0.05) for the tool feed. But, in the case of the nose radius, p = 0.156 (p > 0.1), so it was found to be insignificant. Table 23 shows the results of ANOVA at the factor level decision making and it is concluded that all factors other than the nose radius of the tool are considered.
The table of coefficients is shown in Table 24, from which one can take a deep decision on the degree of the   influence of the factor at its level of input. Similar to Table 23, here, the p-value indicates the significance of the level of factors on the response. Table 24 reflects the same decision  as in Table 20 with some insights into the recommended range of input variation on the cutting velocity of 20-60 m·min −1 and the tool feed of 0.010-0.025 mm·rev −1 . Under unavoidable circumstances, the range may extend up to 0.050. The tool with a nose radius of 0.3 mm can be used for the minimum cutting zone temperature. Hence, the change in the tool radius is not recommended, and it is clear that there is a contribution from the cutting speed and tool feed, and the contribution from the nose radius of the tool used is negligible. The combined effects of the cutting speed and tool feed on the response of the cutting zone temperature are graphically depicted in Figure 21. The validation of a statistical model for the cutting zone temperature response can be observed from the R 2 value of the model summary. Table 25 shows the model summary for the Taguchi analysis on the cutting zone temperature response. The condition is that the R 2 value must be greater than 95% for a good model. In this case, the R 2 value is 99.16%; hence, the model is acceptable. Apart from these, the R 2 values ensured the reliability of the mathematical model as shown in Eq. (4).

Taguchi analysis of the tool wear
Taguchi analysis gives the degree of possibilities to achieve the objective function. Here, the objective is to minimize the tool wear (the smaller the better).
For this statistical model, the intervention group observations of tool wear on 16 different experiments were used to optimize the proposed method. Figure 22 shows the main effect plots for signal-to-noise ratios for the response observations of tool wear during experiments with the proposed nanofluid. The mean line indicates large signals (possibilities towards the objective function of reducing the tool wear). The top leftmost graph for the cutting velocity shows an increase in tool wear with an increase in the cutting velocity; in other words, a decrease of possibilities (signals) when minimizing the tool wear. Hence, the input setting of Level 1, 20 m·min −1 (as the signal-to-noise ratio is 19.748) cutting velocity, 0.025 mm·rev −1 feed rate (as the signal-tonoise ratio is 13.219), and 0.3 mm nose radius of the tool (as the signal-to-noise ratio is 11.735) gave minimum tool wear (refer to Figure 22 and Table 26). The variability is found to be higher in the cutting velocity input than in   the tool feed and lower in the nose radius of the tool used. Table 26 shows the results of the Taguchi analysis, which was the yield of S/N ratios. Figure 23 shows that all observations are statistically accepted as there is good nearness in the mean line in the residual plot. Table 27 illustrates the results of the ANOVA on the response of the tool wear for the decision on the influence of factors. In this stage, one can take a decision on the factor level based on the p-value. Here, the minimum p-value influences are more. The decision criteria are as follows: if p < 0.05, the factor is significantly influencing the measured response, and p > 0.1 indicates an insignificant factor. From Taguchi analysis, it was observed that the cutting velocity is of Rank 1 and its significant value is 0.001 (very low); similarly, p = 0.004 (p < 0.05) for the tool feed. But, in the case of nose radius, p = 0.940 (p > 0.1), so it was found to be insignificant. Table 27 shows the results of ANOVA at the factor level decision making and confirms that other than nose radius, all independent variables considered are significant and they can be used to control the process for rapid and finetuning for minimum tool wear. Figure 24 exhibits the 3D surface plot that shows the relationship between the tool feed and cutting velocity in the tool wear response.
The table of coefficients is shown in Table 28, from which one can take a deep decision on the degree of influence of the factor at its level of input. As in Table 27, the p-value indicates the significance of the level of factors on the response. Though Table 28 gives the same decision, it also gives some deep insights for controlling the tool wear: the cutting velocity should be varied from 20 to 60 m·min −1 , the tool wear from 0.010 to 0.050 mm·rev −1 , and 0.03 mm nose radius should be used than other three types of tools. Table 28 mainly helps to form the mathematical model by generating the regression equation. From the mathematical model, one can interpret the values of untested combinations or predict the process parameter levels for the desired response. The regression model is given as follows:  The validation of the statistical model for the tool wear response can be observed from the R 2 value of the model summary. Table 29 furnishes the model summary for the Taguchi analysis on the tool wear response. The condition is that the R 2 value must be greater than 95% for a good model. In this case, the value of R 2 is 99.89%.   Hence, the model is acceptable. Moreover, the R 2 value also confirmed the reliability and accuracy of the prediction model (Eq. (5)). Hence, the proposed manufacturing process outperformed other methods. The machinability includes a reduction in all five aspects: the cutting force, feed force, cutting zone temperature, surface roughness, and tool wear [115,116]. The limitation of this study is that the influence of specially coated tool inserts was not included in the examinations. The harder tool suffers from less tool wear, and the machinability of the surface roughness will be reduced further. Another limitation is that too small implants could not be machined by using this processing method. The future scope shall include process variables that do not affect the quality of the implant thermally and chemically to improve the surface quality. The tested samples are exhibited in Figure 25.

Scanning electron microscopy examination
Though we have considered the wear rate here, the scanning electron microscopy examination was carried out to observe the nature of the tool wear. The tool wear was observed at a low cutting speed (left side image in Figure 26) and high cutting speed. As the work material is softer compared to the tool material, the shape changes were not significant. But little flank wear was observed.
Venkatesan et al. [52] utilized fresh coconut oil for preparing the nanofluid by mixing 0.25 wt% Al 2 O 3 nanoparticles for machining Inconel 617 by CNC turning. The process parameters, such as cutting speed and feed rate, were optimized based on their impact on cutting force, surface roughness, and tool wear. For their statistical models, the R 2 value was 82.27% for the surface roughness, 78.04% for the cutting force, and 72.31% for the tool wear, and no confirmation experimentation was reported. Yücel et al. [35] prepared the nanofluid by enhancing the commercial conventional nanofluid by mixing 0.6 vol% of MoS 2 for machining the aluminium alloy 2024 T3. The temperature was reduced by 21°C in the MQL mode using a commercial fluid than dry machining and reduced by 43°C using MoS 2 -based nanofluid in the MQL mode. The surface roughness was reduced by 0.432 µm in the MQL mode using a commercial fluid than dry machining and reduced by 0.728 µm using a MoS 2 -based nanofluid in the MQL mode. Şirin and Kivak [40] utilized a concentration of 0.25 vol% for each kind of nanoparticle and maintained a total concentration of 50 vol%. The graphite, MoS 2 , and boron nitride nanoparticles were used in all three possible combinations for preparing the hybrid nanoparticle mixed nanofluid in MQL; for machining the graphite and boron nitride hybrid nanoparticles based nanofluid in MQL, the cutting force was reduced by 10.22 and 3.77%, the peak temperature by 6.92 and 10.78%, and the surface roughness by 14.95 and 8.21%, whereas tool life value improved by 36.17 and 6.08% compared to graphite/MoS 2 and boron nitride.   This research utilized a low concentration of MoS 2 nanoparticles (0.3 wt%) with waste oil (used coconut oil) as a base fluid to prepare the nanofluid for machining biocompatible magnesium implants for biomedical applications under flood cooling conditions. The nanofluid can be reused multiple times in flood cooling, whereas it cannot be reused under MQL conditions. The prepared nanofluid is not harmful to humans as it has edible oil as a base fluid and flood cooling generates less mist than MQL, so it exerts great care for the operator and other workers in the shop. It is a biodegradable coolant with no harm to environmental pollution. The experimental results were compared with the results using a commercial cutting fluid. This research contains two phases that first compared the performance of the proposed nanofluid with a conventional commercial nanofluid in terms of reduction of the cutting force, feed force, cutting zone temperature, tool wear, and surface roughness. Then, the process parameters were optimized for the bestperforming method (the proposed nanofluid with a conventional commercial nanofluid under flood cooling conditions).
This research considered average performance under all cutting conditions for comparison for the recommendation. This improved the reliability of the decision. The proposed nanofluid averagely reduced the cutting force by 68.232 N, the feed force by 34.180 N, the surface roughness by 0.118908 µm, and the tool wear by 0.39938 mg·h −1  compared to the conventional commercial cutting fluid. The cost of cutting is a considerable expenditure in metal machining processes. The regression model was developed based on the experimental results of the cutting force, feed force, cutting zone temperature, surface roughness, and tool wear and presented. The accuracy of developed regression equations for predicting the responses was confirmed as follows. The statistical models were verified with ANOVA. The R 2 values were based on proposed nanofluid observations under flood cooling conditions: 99.22% for the cutting force model, 98.94% for the feed force model, 99.73% for the surface roughness model, 99.16% for the cutting zone temperature model, and 99.89% for the tool wear model. As the obtained R 2 values are greater than 95%, it indicates that there is good agreement between the predicted responses with the use of the regression model and experimental results.
From the Taguchi analysis, it was observed that the nose radius of the tool does not influence the responses significantly, and the best values were obtained at 0.3 mm nose radius. The other two factors were significantly influenced as they possess very low p values [36]. Hence, the cutting speed and tool feed were considered for surface plots. For better presentation, the 3D surface plot was used to reveal the relationship between the above-said variables for responses of cutting force, feed force, cutting zone temperature, tool wear, and surface roughness [37]. The highly influenced parameters were decided based on the experimental results based on Taguchi analysis results [38]. The ranking was based on the influence of the factor on the response. The ranking of variables is shown in Tables 8 and 12 and 17 and 22 and and 26 for the cutting force, feed force, surface roughness, cutting zone temperature, and tool wear responses. The higher delta values indicate a high influence of factors for the concerned response. Venkatesan et al. [52] recommended Taguchi analysis for optimizing the process parameters effectively. They also recommended ANOVA for obtaining the significance of influencing process parameters [39]. The process parameters were optimized with the help of the Taguchi analysis for all five responses and presented [37]. It was found that the cutting velocity and tool feed parameters were highly influenced in all five responses significantly [38]. However, the nose radius of 0.3 mm seems to be the best, as inferred from the Taguchi analysis.
The statistical evaluation showed the significance values p = 0.001, p = 0.023, p = 0.001, p = 0.045, and p = 0.028 for observations of the cutting force, feed force, surface roughness, cutting zone temperature, and tool wear by using the proposed nanofluid and conventional commercial coolant. All these p-values were less than 0.05. The comparison performed in this work was done on average values.

Conclusions
This work demonstrates the low-cost, high-performance cutting fluid (nanofluid) developed from the multiple times used coconut oil (waste) mixed with MoS 2 nanoparticles. The concentration is optimized by the trial and error method. The prepared nanofluid was characterized well before experimentation. The prepared nanofluid was tested experimentally and compared with the performance of the conventional commercial coolant. These observations are statistically significant and confirm that these observations can be accepted. Hence, the use of the proposed nanofluid in place of a conventional commercial coolant averagely reduced the cutting force by 68.23167 N, the feed force by 34.180 N (as the nanofluid supplied sufficient lubrication action for minimizing the forces and absorbing some vibrational effects by maintaining a layer in between the tool and workpiece), the cutting zone temperature by 60.435°C (due to nanofluid lubrication for easy shearing of the material with a hard tool in supplement with cooling), the surface roughness by 0.118908 µm (the lubricated shearing allowed one to make fine cutting and prevented fast wear of sharp tool edges), and the tool wear by 0.39938 mg·h −1 (the lubricated cutting reduced the tool force and permitted the operation to occur smoothly). The proposed method (use of a nanofluid under flood cooling) outperformed other methods. To generalize these outcomes, the process parameters were optimized for using the proposed method for obtaining the best performance. The optimal conditions were presented, and mathematical models (regression equation) were developed to predict the responses. The mathematical models' accuracies were verified with R 2 values. The R 2 values confirmed that the predictions of the developed models have good agreement with experimental results, and the model is accurate. The proposed nano coolant was prepared with extreme care for hightemperature application by concentrating MoS 2 in used coconut oil without the water content by understanding the physical phenomenon of cutting processes. The expected outcomes are given in the experimental results.
Within the limitation of this study, the coolant was tested for machining magnesium samples. In the future, this coolant will be inspected with the machining of precision materials, in CNC milling, CNC grinding, conventional milling, deep drilling processes, etc.
Funding information: The authors state no funding involved.