Study on Machining Parameters Analysis and Optimization for Material Removal Rate and Surface Roughness During Dry Turning of AZ31B Magnesium Alloy Using Ceramic-Coated Carbide Tool Inserts

Abstract

This paper presents research findings on the turning of AZ31B magnesium alloy using ceramic-coated tungsten carbide tool inserts in a dry environment. Fifteen experiments were conducted according to the Box–Behnken design (BBD) for the straight turning of AZ31B magnesium alloy to investigate the variations in two important machinability indicators, i.e., material removal rate ‘MRR’ and mean roughness depth ‘R_Z’, with variations in cutting speed ‘C_S’, feed rate ‘f_r’, and depth of cut ‘DoC’. The cutting speed and feed rate had the maximum influence on the mean roughness depth and material removal rate, respectively. To address the challenge of optimizing conflicting machining responses, desirability function analysis (DFA) and grey relational analysis (GRA) were employed to identify the optimal turning parameters for conflicting machinability indicators or responses. These techniques enabled the simultaneous maximization of the material removal rate and the minimization of the mean roughness depth, ensuring an effective balance between productivity and surface quality. The optimal turning conditions—cutting speed of 90 m/min, feed rate of 0.2 mm/rev, and depth of cut of 1.0 mm—yielded the best multiperformance results with an MRR of 18,000 mm³/min and an R_Z of 2.21 µm. Scanning electron microscope (SEM) analysis of the chip and flank surface of the cutting tool insert used in the confirmation tests revealed the formation of band-saw-type continuous chips and tool wear caused by adhesion and abrasion.

1. Introduction

Magnesium (Mg) is considered one of the leading green materials of the 21st century [1]. Magnesium is the third most widely used structural element after steel and aluminum. Magnesium (Mg) is a lightweight, silvery-white metal with about two-thirds the density of aluminum. It has a low melting point, good electrical conductivity, moderate thermal conductivity, and high chemical reactivity. Magnesium alloy is a metal that is made by combining magnesium with other metals, such as aluminum, zinc, or manganese, to improve its properties [2,3]. Some key characteristics of magnesium alloys include being lightweight, having a high strength-to-weight ratio, corrosion resistance, good machinability, and poor fatigue resistance. Magnesium alloys are important materials for various engineering applications, such as automotive, aerospace, and electronics. Common magnesium alloys include the AZ series (Al-Zn), AM series (Al-Mn), and ZM series (Zn–rare earth) [4].

AZ31B is abundantly available, and its high strength-to-weight ratio and low density make it ideal for a variety of applications, including orthopedic implants, aircraft fuselages, and electronic products [5,6]. AZ31B magnesium alloys undergo extensive machining during product manufacturing, presenting several challenges. The major machining issues include flammability (fire risk), chip formation (clogging and tool wear), rapid tool wear (due to abrasiveness), low thermal conductivity (leading to overheating and poor surface finish), surface defects (burr formation), and work hardening (increasing cutting forces and reducing efficiency). Additionally, at high temperatures, the material is prone to chipping and adhering to tool faces. To mitigate these issues, appropriate actions are needed, such as (i) optimizing machining parameters like cutting speed, feed rate, depth of cut, and coolant application; (ii) using suitable tools with enhanced cutting ability, such as coated or treated carbide or ceramic cutting tools; and (iii) applying specific lubricants or coolants to prevent overheating and improve surface quality [7,8].

Several conventional machining methods are used to machine magnesium alloys in both wet and dry environments. However, many processes that rely on coolants (wet environments) are unsustainable and have detrimental effects on the environment. Using coolants in machining processes introduces several negative impacts, including environmental pollution, health and safety risks, and corrosion. Additionally, issues such as improper chip disposal and health hazards from contaminated fluids arise. The need for increased maintenance, including cleaning and filtration to prevent clogs, contamination, and system failures, further contributes to operational challenges. Ultimately, these factors lead to higher overall machining costs. This drives researchers to explore sustainable machining methods for soft materials, such as magnesium, intending to achieve improved part quality, enhanced sustainability, and increased process productivity. Dry machining is one such technique that eliminates the need for lubrication, significantly reducing machining costs while helping to maintain a cleaner and greener environment by minimizing emissions [9,10]. Dry machining is the process of machining materials without cutting fluids, such as coolants or lubricants. It reduces environmental impact and operational costs by eliminating coolant disposal, filtration, and maintenance. This method also enhances safety by reducing exposure to harmful fluids and helps to prevent corrosion and contamination. Dry machining offers several advantages; however, it is often inadequate for achieving optimal machinability, especially with challenging materials or complex processes. Successful outcomes require integrating additional techniques, such as specialized tools and optimized machining parameters. For instance, coated cutting tools can enhance tool life and performance, but appropriate combinations of machining parameters may be needed [11,12].

Ceramic-coated tool inserts are cutting tools coated with a thin layer of ceramic material, such as alumina (Al₂O₃), titanium carbonitride (TiCN), or other hard ceramics, to enhance their performance in machining processes. These tool inserts provide numerous benefits, including enhanced wear resistance for a longer tool life, especially when machining abrasive materials. Their thermal stability allows them to perform at high temperatures, making them ideal for high-speed cutting. The coating also offers oxidation resistance, reduces friction, improves chip flow, and enhances the surface finish. These inserts support higher cutting speeds, improve productivity, and are corrosion-resistant, making them suitable for harsh environments. Their versatility allows the machining of various materials, including tough alloys, making them a top choice for precision, durability, and performance in demanding operations [13,14]. Some important past work on magnesium alloy machining is discussed below.

Deswal and Kant performed laser heat-assisted dry turning of an AZ31B-grade magnesium alloy [15]. They compared the results with the outcomes of plain turning and found a significant enhancement in machinability, in terms of obtaining a lower roughness, machining force, and tool wear, with laser assistance. Carou et al. [16] carried out the dry turning of UNS M11917 magnesium alloy at varying turning combinations and found feed rate ‘f_r’ to be the most influencing parameter on surface roughness. They reported keeping the maximum temperature below 50 °C and setting the machining parameters accordingly for better machinability prospects. Rubio et al. [17,18] used tool inserts of varying coatings to face-turn magnesium alloy in the absence of lubricants. They obtained a high productivity at high values of cutting parameters. For the lowest surface roughness, they obtained a 0.04 mm/rev feed rate, 280 rpm spindle speed, and a TP2500 coating grade as the optimum combination of process parameters. Guo et al. [19] found a significant influence of the depth of cut and feed rate on the machinability of AZ91D alloy. A recent study on the turning of AZ91 magnesium alloy reported the successful minimization of cutting forces [20]. It was recommended to prevent the adhering of chips to the tool face and avoid machining at high speed. A previous article [21] presented an investigation on the wear of the tool inserts and the morphology of the chips while cutting AZ31B magnesium in a dry environment under the same set of conditions reported in the present article.

The literature review indicates that magnesium is a soft and flammable material and, therefore, is prone to ignition. In a dry environment, this may increase the machining risks. To obtain the best machinability indicators, an optimum combination of parameters is desirable. Moreover, previous attempts at dry machining AZ31B magnesium alloys are scarce. The review of past studies reveals several key research gaps, including (i) the limited application of multi-response optimization techniques, particularly for the simultaneous optimization of conflicting responses such as the material removal rate and mean roughness depth; (ii) limited details on the usage of ceramic-coated tool inserts in dry turning operations; (iii) a predominant focus on average surface roughness, with minimal consideration of other better and important surface quality indicators such as mean roughness depth ‘R_Z’; (iv) a lack of comparative analysis between optimization methods like desirability function analysis (DFA) and grey relational analysis (GRA); and (v) limited validation of findings through confirmation experiments. The present study attempts to fill the gap and explores the dry turning of the AZ31B magnesium alloy using a thin film of ceramic-coated carbide tool inserts. The primary goal is to enhance the machinability of AZ31B magnesium alloy by maximizing the material removal rate (MRR) and minimizing the mean roughness depth (R_Z). The specific objectives of this research are (i) to explore the sustainable straight turning of AZ31B alloy under dry conditions, ensuring minimal tool wear and eliminating fire risks; (ii) to analyze the effects of three key machining parameters—cutting speed (C_S), feed rate (f_r), and depth of cut (DoC)—i.e., the mean roughness depth ‘R_Z’ and productivity (i.e., MRR); (iii) to apply multi-response optimization techniques (DFA and GRA) to identify optimal turning parameters for simultaneously maximizing productivity (MRR) and minimizing surface roughness (R_Z); (iv) to validate the optimized results through confirmation tests; and (v) to examine the tool flank wear and chip morphology obtained during the confirmation experiments.

This study contributes to the scientific understanding of the dry machining of AZ31B magnesium alloy by systematically applying advanced multi-response optimization techniques (DFA and GRA) to simultaneously enhance surface quality and productivity. By identifying the optimal combination of turning parameters for maximizing MRR and minimizing R_Z, this research offers a validated framework for sustainable and efficient machining. Additionally, it provides new insights into tool wear behaviour and chip morphology under dry conditions using ceramic-coated carbide inserts.

2. Experimental Details

2.1. Material

In this study, experiments were conducted on a Colchester manual lathe machine (manufacturer: Colchester Machine Tool Solutions; model: Mascot 1600, city and country: West Yorkshire, United Kingdom) performing straight turning with thin ceramic-film-coated tungsten carbide tool inserts under dry conditions. The work material used was round bars of AZ31B magnesium alloy. A total of eight bars, each with a diameter of 20 mm and a length of 200 mm, were employed for the experimentation. The chemical composition (in % wt.) of the AZ31B magnesium alloy bars was as follows: Mg—97%, Al—2.5%, Zn—0.6%, Mn—0.2%, Si—0.1%, Cu—0.05%, Ca—0.04%, Fe—0.005%, and Ni—0.005%. Sieeso (TNMG160408-MT HS8125) (manufacturer: Zhuzhou Sieeso Advanced Materials Corporation Limited; city and country: Zhuzhou, China) carbide tool inserts, coated with a thin alumina ceramic film, were used for the straight turning of AZ31B magnesium alloy bars under dry conditions. The inserts featured a triangular (T) shape with negative (N) geometry and multi-use (M) characteristics, offering three effective cutting edges (single-point), as shown in Figure 1. These inserts were coated with medium-thickness titanium carbonitride (TiCN) and a thin layer of alumina (Al₂O₃) with a combined ceramic coating hardness of approximately 3000 HV. This coating enhances performance in high-speed and high-temperature machining applications. It improves wear resistance and toughness, allowing the insert to withstand high cutting temperatures and to extend the tool life, especially in heavy-duty operations. These inserts are widely used in turning operations due to their durability and performance under demanding conditions, offering a long tool life, excellent thermal stability, and oxidation resistance. Their versatility makes them suitable for machining a wide range of materials. Figure 1 shows the three-dimensional views of the cylindrical AZ31B magnesium alloy bar and ceramic-coated tool insert.

2.2. Experimentation and Measurement

A Colchester Mascot 1600 manual lathe machine (12.5 HP, 1600 RPM) was used for the straight turning of AZ31B magnesium alloy bars with ceramic-coated carbide tool inserts, as shown in Figure 2. Fifteen machining combinations were designed using the Box–Behnken Design (BBD), as detailed in

Table 1. Experimental combinations for turning of Mg alloy (AZ31B) and corresponding values of MRR and R_Z with grey relational grade.

Ex. No.	Variable Machining Parameters			Machining Performance Indicators						Grey Relational Grade (GRG)
				MRR (mm3/min)			Mean Roughness Depth ‘RZ’ (µm)
	Cutting Speed ‘CS’ (m/min)	Feed ‘f’ (mm/rev)	Depth of Cut (mm) ‘DoC’ (mm)	R1	R2	Average (R1 + R2)	R1	R2	Average (R1 + R2)
1	90	0.15	0.75	10,125	10,125	10,125	1.98	2.32	2.15	0.59
2	65	0.20	0.75	9750	9750	9750	2.26	2.58	2.42	0.55
3	115	0.15	0.50	8625	8625	8625	4.23	3.99	4.11	0.40
4	90	0.10	1.00	9000	9000	9000	2.36	2.28	2.32	0.55
5	65	0.15	1.00	9750	9750	9750	1.81	1.63	1.72	0.68
6	65	0.10	0.75	4875	4875	4875	1.62	1.52	1.58	0.67
7	90	0.20	0.50	9000	9000	9000	2.29	2.03	2.16	0.58
8	115	0.15	1.00	17,250	17,250	17,250	4.81	4.47	4.64	0.62
9	90	0.20	1.00	18,000	18,000	18,000	2.45	2.27	2.36	0.83
10	90	0.10	0.50	4500	4500	4500	1.92	1.74	1.83	0.59
11	90	0.15	0.75	10,125	10,125	10,125	2.31	2.07	2.19	0.59
12	65	0.15	0.50	4875	4875	4875	1.67	1.47	1.57	0.67
13	115	0.20	0.75	17,250	17,250	17,250	4.61	4.31	4.46	0.62
14	115	0.10	0.75	8625	8625	8625	2.67	2.99	2.83	0.48
15	90	0.15	0.75	10,125	10,125	10,125	2.31	2.11	2.21	0.58

. The selected machining parameters included cutting speeds (C_S) at 65, 90, and 115 m/min; feed rates (f_r) at 0.1, 0.15, and 0.2 mm/rev; and depths of cut (DoC) at 0.5, 0.75, and 1.0 mm [21]. Each experiment was conducted for 15 min and repeated twice. The material removal rate (MRR) and surface roughness, i.e., mean roughness depth ‘R_Z’, were considered two important indicators for evaluating the machinability of AZ31B magnesium alloy. MRR represents the volume of material removed in the form of chips. The following equation was used to calculate the MRR:

$$MRR=\mathrm{C}\mathrm{u}\mathrm{t}\mathrm{t}\mathrm{i}\mathrm{n}\mathrm{g}\hspace{0.17em}\mathrm{s}\mathrm{p}\mathrm{e}\mathrm{e}\mathrm{d}\hspace{0.17em}\mathrm{F}\mathrm{e}\mathrm{e}\mathrm{d}\hspace{0.17em}\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{e}\times \mathrm{D}\mathrm{e}\mathrm{p}\mathrm{t}\mathrm{h}\hspace{0.17em}\mathrm{o}\mathrm{f}\hspace{0.17em}\mathrm{c}\mathrm{u}\mathrm{t}$$

(1)

The surface roughness value presents the amount of irregularity in the machined surface and is responsible for its performance while in use [22]. The mean roughness depth ‘R_Z’ was used to determine the surface quality. Surface roughness was measured using the ART300 portable tester (manufacturer: AJR-NDT Corporation Limited; city and country: Xingtai, China). Measurements were taken with a 5 µm diameter stylus, using a Gaussian filter with a cut-off length of 0.8 mm and an evaluation length of 2 mm. Three measurements were taken across the machining direction on the turned magnesium bar, and their average values were taken into consideration. For tool wear measurements, ISO recommendations were followed and a 600 µm threshold of flank wear was considered as the end of tool life. Both chip morphology and tool flank wear were investigated on the TESCAN VEGA3 scanning electron microscope ‘SEM’ (manufacturer: TESCAN; city and country: Brno, Czech Republic). Figure 2 presents the experimental setup used in the present work.

3. Results and Discussion

The combinations of variable turning parameters for the manual lathe tool, designed based on the BBD-RSM with fifteen experimental runs, along with the average values of the measured responses from two replicates for each run, are presented in Table 1. Statistical testing of the model terms is conducted using analysis of variance (ANOVA), as shown in

Table 2. ANOVA results of responses selected for turning AZ31B magnesium alloy bars with a manual lathe using ceramic-coated carbide inserts.

For material removal rate ‘MRR’ (mm3/min)
Source	SS	DF	MS	F	p	Remark
Model	2.455 × 108	3	8.184 × 107	74.44	<0.0001	Significant
CS	6.328 × 107	1	6.328 × 107	57.56	<0.0001	Significant
fr	9.112 × 107	1	9.112 × 107	82.88	<0.0001	Significant
DoC	9.113 × 107	1	9.113 × 107	82.88	<0.0001	Significant
Residual	1.209 × 107	11	1.099 × 106
Lack of fit	1.209 × 107	9	1.344 × 106
Pure error	0.000	2	0.000
Cor total	2.576 × 108	14
R-Squared = 0.951, Adjusted R-Squared = 0.9503, Predicted R-Squared = 0.8995
PRESS = 2.590 × 107, Adequate Precision = 24.932
For mean roughness depth ‘RZ’ (µm)
Model	13.15	9	1.461	6.40	0.027	Significant
CS	0.27	1	0.266	1.17	0.3294	Significant
fr	0.031	1	0.031	0.14	0.7272
DoC	0.016	1	0.016	0.07	0.8027
CS fr	0.16	1	0.156	0.68	0.4461
CS DoC	0.036	1	0.036	0.16	0.7073
fr DoC	0.021	1	0.021	0.09	0.7738
(CS2)	2.03	1	2.026	8.87	0.0308	Significant
(fr2)	0.038	1	0.038	0.17	0.6996
(DoC2)	0.027	1	0.027	0.12	0.7440
Residual	1.14	5	0.228
Lack of fit	1.14	3	0.380	407.10	0.0025	Significant
Pure error	1.87 × 103	2	0.001
Cor total	14.29	14
R-Squared = 0.9201, Adjusted R-Squared = 0.7763, Predicted R-Squared = −0.2762
PRESS = 18.24, Adequate Precision = 7.433

. ANOVA is conducted using a 95% confidence interval (p < 0.05) to assess the significance of the developed models, turning parameters, and their interactions for each response. The results of the analysis of variance (ANOVA), as presented in Table 2, yield the following interpretations: the ANOVA test is a key statistical tool for assessing data fit, model adequacy, and identifying influential parameters [23].

• The developed quadratic model and linear model for MRR and R_Z, respectively, are significant because the p-values are less than 0.05, as per the 95% confidence interval.
• Only C_S is found to be statistically significant for R_Z. However, all variables are found to be statistically significant for MRR.
• The square term of cutting speed is found to be significant for the mean roughness depth ‘R_Z’.
• The R-squared values of the developed MRR and R_Z models are close to 1, confirming their strong predictive accuracy and reliability.
• Adequate precision values above 4 indicate a desirable signal-to-noise ratio. The MRR and Rz models show values of 24.93 and 7.26, respectively, confirming that the developed models have adequate signals and are suitable for prediction.
• The predicted R-squared value for the MRR model is in good agreement with the adjusted R-squared (difference < 0.2), indicating a strong correlation between the experimental and predicted values.
• A normal distribution of the data is confirmed in Figure 3, with all fifteen residuals closely aligning with the mean line. Measured data points are represented in different colours based on their value ranges. In Figure 3a,b, red, green, and blue data points indicate high, intermediate, and low values of MRR and R_Z, respectively.
• Empirical Equations (1) and (2) are prediction models for MRR and R_Z, respectively.

The empirical equations derived for the responses are as follows:

(i) For the material removal rate ‘MRR’:

$$MRR\left({\displaystyle \frac{\mathrm{m}{\mathrm{m}}^3}{\mathrm{m}\mathrm{i}\mathrm{n}}}\right)=-2050+112.5\hspace{0.17em}{C}_S+\mathrm{67,500}\hspace{0.17em}{f}_r+\mathrm{13,500}\hspace{0.17em}DoC$$

(2)

(ii) The equation for the mean roughness depth ‘R_Z’:

$$\begin{gathered} R_z\left(\mu \mathrm{m}\right)=+8.63-0.205C_S+9.43f_r-1.873DoC+0.158C_S{f}_r+0.0152C_{S}DoC-5.8f_{r}DoC \\ +0.0012C_S^2-40.67f_r^2+1.373Do{C}^2 \end{gathered}$$

(3)

Empirical Equations (1) and (2) are the developed models for predicting the material removal rate (MRR) and mean roughness depth (R_Z), respectively, based on the various combinations of turning parameters listed in Table 1. The predicted values of MRR and R_Z are compared with the corresponding experimental values in Table 1 to validate the accuracy of the models. Figure 4 presents the variation in the experimental and predicted values of the considered responses across the experimental runs. Figure 4a,b compare the material removal rate (MRR) and mean roughness depth (Rz), respectively, with the blue and orange lines indicating the experimental and predicted values. The vertical axis shows the response values, while the horizontal axis denotes the experimental runs, each representing a unique set of turning parameters. A close agreement between the experimental and predicted results confirms the reliability of the models. These results indicate that empirical Equations (2) and (3) demonstrate strong predictive accuracy, as evidenced by the close agreement between the experimental and predicted values of MRR and R_Z. This indicates that the models effectively capture the influence of turning parameters on the responses. However, the models are limited to the specific range of parameters used in the experiments and may not generalize well beyond these conditions. Additionally, they do not account for potential nonlinear interactions or external factors such as tool wear or material inconsistencies.

Influence of Variable Turning Parameters on Responses

Figure 5a,b illustrate the variation in MRR and R_Z with three selected turning parameters, i.e., C_s, f_r, and DoC. The coded values of the variable turning parameters are on the abscissa, whereas the actual values of the machinability indicators, i.e., MRR and R_Z, are on the ordinates. The variation in MRR with DoC is depicted on the secondary axis in Figure 5a, while the variation in MRR with C_S and f_r is depicted on the primary axis. It is evident from Figure 5a that MRR increases with an increase in the turning parameters. Machining at high C_S, f_r, and DoC results in the maximum MRR. An increased spindle speed, movement of the tool along the workpiece, and the insertion of the tool at higher depths and at high values of turning parameters cause the removal of higher amounts of work materials and chips and, hence, increase the material removal rate. Figure 5b shows that the mean roughness depth ‘R_Z’ increases with C_S, f_r, and DoC. C_S is the most influential on R_Z. R_Z increases linearly with the cutting speed and varies rapidly in the 90–115 m/min speed zone. Turning of the magnesium alloy at high speed causes the generation of heat and thereby causes the softening of the tool and the further chipping of the work particles. Due to magnesium being adhered on the tool face, continuous machining deteriorates the turned work surfaces further, which is reflected in the high surface roughness. Excessive heat generation due to high friction, work bar wobbling, and vibration and chattering corresponding to high values of cutting parameters results in increased surface roughness.

4. Optimization

Finding the optimal combination of variable parameters is a critical task in any machining operation, as it aims to achieve both high surface quality and productivity simultaneously [24,25]. Desirability function analysis (DFA)- and grey relational analysis (GRA)-based techniques are used for optimization to maximize the material removal rate ‘MRR’ and minimize the mean roughness depth ‘R_Z’ simultaneously. Both DFA and GRA are valuable for multi-response optimization due to their complementary approaches. DFA is a goal-oriented technique that prioritizes responses using weights [26], while GRA is a relation-based method which focuses on closeness to ideal values without needing predefined preferences [27]. Their optimization results may differ slightly, offering broader insight and more reliable decision-making when used together. DFA is particularly well suited for optimizing conflicting responses simultaneously, as it allows the assigning of weights and importance levels to each response, helping to balance trade-offs based on specific goals. GRA is best suited when responses have similar goals or importance, as it treats all responses uniformly by measuring their closeness to the ideal solution without requiring predefined weights. DFA identifies the exact optimum values within a specified range of variable parameters, although these values may not correspond exactly to the levels used in the experimental design. However, they may align with the available machine’s incremental values or be close to them. In contrast, GRA-based optimization typically results in parameter values that closely match one of the experimental runs defined in the design, making it more directly applicable to the experimental setup.

In this study, the objective function is the larger-the-better type and the smaller-the-better type for the material removal rate and mean roughness depth, respectively. Therefore, Equations (4) and (5) are used to compute the desirability of MRR and R_Z for the ith experimental observation and a target value equal to a maximum value of MRR and a minimum value of R_Z identified from experimental values, as given in Table 1, is used.

For the maximization of the material removal rate (i.e., larger-the-better type)

$$d_j={\left({\displaystyle \frac{y_j-L}{O-L}}\right)}^{w_j}$$

(4)

For the minimization of the mean roughness depth (i.e., smaller-the-better type)

$$d_j={\left({\displaystyle \frac{H-y_j}{H-O}}\right)}^{w_j}$$

(5)

$$C_d={\left(d_1,d_2,d_3\dots \dots \dots \dots .d_t\right)}^{{\displaystyle \frac{1}{t}}}={\left(\prod _{j=1}^t{d}_j\right)}^{{\displaystyle \frac{1}{\sum W_j}}}$$

(6)

where ‘d_j’ is the desirability of all machining performance indicators ‘y_j’; ‘C_d’ is the collective desirability function; O is the objective value; H and L are the permissible upper and lower values of the ith response, respectively; w_i is weightage allocated to the response y_i; and t is the total number of responses. To achieve the best possible combination, a collective desirability (C_d) value that is closer to 1 is chosen. Equations (7) and (8) are used to compute the desirability for the jth combination of MRR and R_Z by assigning an equal weightage of 0.5.

For the maximization of MRR (i.e., larger-the-better type)

$${d_{{MRR}_J}=\left({\displaystyle \frac{{MRR}_J-{MRR}_{min}}{{MRR}_{max}-{MRR}_{min}}}\right)}^{0.5}$$

(7)

For the minimization of the R_Z (i.e., smaller-the-better type)

$${d_{R_{Z_j}}=\left({\displaystyle \frac{R_{Z_{max}}-R_{Z_j}}{R_{Z_{max}}-R_{Z_{min}}}}\right)}^{0.5}$$

(8)

The maximum MRR and R_Z values used are 4.64 µm and 18,000 mm³/min, respectively, while the corresponding minimum values are 1.57 µm and 4500 mm³/min, respectively.

The following Equations (9) and (10) are used to compute the collective desirability function ‘C_dj’ for the jth combination of turning parameters.

$$D_j={\left[\left\{{\left(d_{MRR}\right)}_j^{0.5}{\left(d_{R_Z}\right)}_j^{0.5}\right\}\right]}^{{\displaystyle \frac{1}{0.5+0.5}}}$$

(9)

$$D_j=\left[\left\{{\left(d_{F_W}\right)}_j^{0.5}{\left(d_{R_Z}\right)}_j^{0.5}\right\}\right]$$

(10)

The optimum results obtained from DFA are a 77.39 m/min cutting speed, a 0.2 mm/rev feed rate, a 1.0 mm depth-of-cut, a 15,451 mm³/min MRR, and a 2.11 µm R_Z. Since the optimum value of the cutting speed obtained from the DFA is different from the standard values available in the manual lathe machine tool, the adjacent standard values closest to the optimum are chosen for the confirmation runs, as shown in

Table 3. Results of DFA- and GRA-based optimization and confirmation experiments.

Turning Details		Optimized Value		Confirmation Results		Error (%)
		DFA	GRA	Avg. (CEl1 & E2)
				DFA	GRA	DFA	GRA
Variable turning parameters	Cutting speed ‘CS’ (m/min)	77.39	90
	Feed rate ‘fr’ (mm/rev)	0.2	0.2
	Depth-of-cut ‘DoC’ (mm)	1.0	1.0
Performance indicators	Material removal rate ‘MRR’ (mm3/min)	15,451	18,000	18,000	18,000	16.49	0
	Mean roughness depth ‘RZ’ (µm)	2.11	2.36	2.26	2.21	7.1	6.4

.

In grey relational analysis (GRA)-based optimization, Equations (11) and (12) are used for the objective functions (i.e., maximizing and minimizing) after normalizing the experimental results.

For maximizing MRR (higher-the-better)

$${MRR}_{ij}={\displaystyle \frac{Y_{jk}-\mathit{min}\left(Y_j\right)}{\mathit{max}\left(Y_j\right)-\mathit{min}\left(Y_j\right)}}$$

(11)

For minimizing R_Z (lower-the-better)

$${R_Z}_{ij}={\displaystyle \frac{\mathit{max}\left(Y_j\right)-Y_{jk}}{\mathit{max}\left(Y_j\right)-\mathit{min}\left(Y_j\right)}}$$

(12)

where max (Y_j) and min (Y_j) are the maximum and minimum values of jth performance indicators among the fifteen experiments shown in Table 2. Equation (5) is also used to calculate the grey coefficient ‘GC_jk’ values for the jth response of the kth experiment.

$${GC}_{jk}={\displaystyle \frac{{\mathit{Delta}}_{\mathit{min}}+\gamma {\mathit{Delta}}_{\mathit{max}}}{{Delta}_{jk}+\gamma {Delta}_{max}}}$$

(13)

where ‘$\gamma$’ denotes the distinguishing coefficient, which has a value range of 0 to 1. It is usually allocated to 0.5 to give each machining performance indicator equal weightage. By considering the average values of ‘GC_jk’ of all of the machining performance indicators corresponding to the kth experimental run, Equation (6) is used to compute the grey relational grade (G_g). In Equation (6), p indicates the total number of performance indicators.

$${G_g}_j=\sum _{k=1}^p\left({GC}_{jk}\right)$$

(14)

In GRA, the highest value of G_g indicates the optimal combination. Therefore, experimental run 9 has the highest GRG of 0.83 and is regarded as the best turning combination, consisting of 90 m/min C_S, 0.2 mm/rev f_r, and 1.0 mm DoC (i.e., run 9).

Since the optimum cutting speed obtained from the DFA differs from the cutting speed levels listed in the experimental table and the standard values available in the manual lathe tool, the nearest standard value is selected for the confirmation runs, as shown in Table 3. After adjusting to the closest standard cutting speed, the same turning combination is obtained from both DFA and GRA.

The optimum results obtained from the best desirability (i.e., 0.99) and GRG (0.83) are validated by conducting two confirmation runs (CEs) using an optimal turning combination. The average values of the machining performance indicators obtained from the confirmation runs are used. Table 3 shows the optimum results, as well as the values of MRR and R_Z obtained from the confirmation runs. The average (i.e., CEs 1 and 2) MRR and R_Z values for DFA are 15,451 mm³/min and 2.36 µm, respectively, and for GRA, they are 18,000 mm³/min and 2.21 µm, respectively. The percentage difference between the optimal and confirmation results is under 10%, indicating a good agreement between the predicted and actual values. The higher MRR percentage difference for DFA, shown in Table 3, is attributed to significant variations in cutting speed values during the confirmation test, resulting from the selection of the adjacent standard cutting speed values available on the manual lathe machine.

Analysis of Chip Morphology and Tool Wear

Chip formation in machining occurs when a material is removed from a workpiece, influenced by the cutting speed, feed rate, depth of cut, and the materials of the tool and workpiece [28,29], whereas tool wear in turning is the gradual deterioration of the cutting tool due to friction, heat, and mechanical stress. It can occur as abrasive, adhesive, thermal, chemical, or notch wear, impacting the surface finish, accuracy, and machining efficiency. Monitoring wear and optimizing cutting conditions can help to reduce its effects. SEM analysis of the removed metal from the workpiece and tool insert flank face is conducted. Figure 6a shows the formation of band-saw-type continuous chips under optimal parameters. AZ31B, being a ductile material, deforms plastically under cutting forces, causing chips to stretch and curl into long, continuous forms. Moderate cutting speeds promote this behaviour, favouring plastic deformation over fragmentation. Continuous chips cause abrasive and adhesive wear, with severe magnesium adhesion on the tool flank. In contrast, small, discontinuous chips lead to adhesion or chipping, with major wear due to abrasion. Figure 6b presents scanning electron micrographs of tool insert wear, showing magnesium particle adhesion on the tool flank and wear from both adhesion and abrasion (Figure 6b,c). Ceramic coatings significantly enhance tool performance by severely mitigating the wear mechanisms of abrasion, adhesion, and diffusion. The high hardness and thermal stability of ceramic-coated tools improves their resistance to abrasive wear, while their low chemical reactivity and smooth surface help to reduce adhesive wear. Additionally, ceramic coatings serve as effective barriers against heat and chemical interactions, thereby minimizing diffusion wear. Compared to uncoated inserts, these properties lead to extended tool life, improved cutting stability, and increased overall machining efficiency.

5. Conclusions

This study has investigated the dry cutting of AZ31B magnesium alloy on a manual lathe using ceramic-coated carbide tool inserts. The analysis and effects of the machining parameters and their optimization for surface roughness and material removal rate have been reported. The summary of this research is as follows:

• Successful fire-ignition-free machining was achieved, achieving a higher material removal rate without compromising the surface finish.
• Cutting speed was found to have the most impact on mean roughness depth, whereas feed rate was more prominent for material removal rate.
• MRR and R_Z both increased with an increase in variable turning parameters.
• A 90 m/min Cs, 0.2 mm/rev f_r, and 0.1 mm DoC were obtained as the optimum combination/settings of the machining parameters from GRA for the best surface quality, with a mean roughness depth R_Z of 2.21 µm and a best productivity with MRR of 18,000 mm³/min.
• The SEM study revealed the formation of band-saw-type continuous chips and flank wear due to adhesion and abrasion with the chipping of magnesium.
• We recommend intermediate Cs (90 m/min) and higher values of f_r (0.2 mm/rev) and DoC (1 mm) to obtain maximum productivity with a better surface finish for AZ31B magnesium alloy.

This study on the dry turning of AZ31B magnesium alloy bars had limitations, such as the use of a controlled set of turning parameters and the use of specific tool materials and coatings, which may have limited the generalizability of the findings to other machining conditions or tool types. While the dry environment offered sustainability benefits, it may also have resulted in a reduced tool life and surface quality compared to lubricated processes. Additionally, the absence of real-time monitoring and thermal analysis limited the ability to fully understand tool wear and material behaviour. These limitations opened up opportunities for future research, which could include (i) the machining of AZ31B magnesium alloy in dry environments using textured or treated tools; (ii) a comparative study of magnesium machining under different cutting environments; (iii) an investigation on the machining of AZ31B magnesium alloy in dry conditions using a wider range of tool materials and coatings, especially those designed for the high-performance machining of light alloys; (iv) utilizing advanced monitoring techniques (e.g., force, temperature, acoustic signals) to asses real-time tool wear and thermal effects; (v) an investigation on surface integrity and subsurface damage, including residual stresses, microhardness, and corrosion resistance, to assess the quality and performance of the magnesium alloy parts; and (vi) utilizing machine learning (ML) and finite element analysis (FEM) to develop predictive models for optimizing machining parameters and enhancing process efficiency and sustainability.