Multi-objective optimization of surface roughness and MRR in AISI 316L stainless steel processed by MQL end milling using taguchi, RSM, ANN, and RFR methods

Abstract

This research improves the cutting parameters for end milling AISI 316L stainless steel, a material that is utilized in a variety of sectors, including nuclear power, food, medicine, chemicals, and the marine sector. It has remarkable corrosion resistance. Its great mechanical qualities and limited heat conduction make it challenging to manufacture. When milling with neem oil under Minimum Quantity Lubrication (MQL), the Taguchi technique was utilized to choose the cutting parameters, with an emphasis on Surface Roughness (Ra) and Material Removal Rate (MRR). Important factors such as feed rates, cutting speeds, and cut depths were examined, as well as morphological changes and chip formation. Tool dynamometers were used to quantify MRR, and a surface finish tester was used to evaluate surface roughness. The cutting parameters were optimized and validated using advanced optimization techniques such as Random Forest Regression (RFR), Back Propagation Artificial Neural Network (BPANN), Feed Forward Artificial Neural Network (FFANN), Desirability Function Analysis (DFA), Taguchi Design of Experiments (TDOE), and Response Surface Methodology (RSM). The findings show that machining efficiency is greatly impacted by Material Removal Rate (MRR). While MQL utilizes a prepared Neem oil enhanced tool life and surface quality, higher cutting speeds, feed velocities, and depths of cut increased MRR. At 150 m/min cutting speed, 250 mm/min feed velocity, and 2 mm depth of cut, the best MRR was obtained. At moderate feed velocities, shallow cuts, and medium cutting speeds (100 m/min), surface roughness was reduced. MRR and surface roughness were successfully predicted by the RSM, BPANN, FFANN, and RFR models; RFR proved to be the most accurate.

Introduction

Stainless steels are used extensively in the food, chemical, health, nuclear, and maritime industries because of their excellent mechanical qualities, low heat conductivity, and corrosion resistance. Custom 450, a martensitic stainless steel, exhibits excellent resistance to pitting and rust in saltwater (up to 650 °C), and heat treatment can further improve it^1,2,3,4. There aren’t many studies on its machinability, though. Because of their high strength and low heat conductivity, stainless steels are challenging to machine, which can result in chip adhesion, surface imperfections, and quick tool wear^3,5,6,7. Feed rate, tool radius, depth of cut, and cutting speed are some of the variables that affect machinability; higher speeds decrease forces, but higher feed and depth increase roughness^8,9,10. Performance is additionally influenced by material, tool geometry, coatings, and cryogenic cooling^{11,12,13,14,15}. Tool notching frequently occurs during the milling of 316L stainless steel^16,17, and smaller insert radii hasten wear¹⁸. Cutting fluids also affect surface quality and tool life^19,20. Cutting speed had a significant impact on surface roughness, whereas coating type had an impact on cutting forces, according to Ciftci’s study of dry turning AISI 304 and 316 ¹². While Selaimia et al. demonstrated that feed rate affected both forces and roughness in milling X2CrNi18-9, Basmaci et al. examined feed rate, depth of cut, and tool tip radius in turning 17-4 PH stainless steel^21,22. Microcracks and coating loss during dry milling of AISI 304 were reported by Varghese et al.¹⁷.

Kuram and Ozcelik used the Taguchi method to find that in micromilling AISI 304, roughness, wear, and forces were considerably impacted by spindle speed, feed rate, and depth of cut²³. By examining the effects of depth, feed, and speed on burrs and surface roughness, Lin enhanced machinability^24,25. Rake angle has a significant impact on tool life when milling Co–Cr steel with TiCN/TiN-coated tools, according to Shao et al.²⁶. In milling STAVAX steel, Liew found that hardness increased wear, and coated tools demonstrated improved resistance to fracture^27,28. When milling austenitic stainless steel, Nordin et al. found that fracture formation was the primary wear mechanism caused by multilayer carbide coatings²⁹. In milling AISI 303 and 304, Selinder et al. showed that their novel coating worked better than traditional PVD and CVD coatings³⁰. In machining martensitic stainless steels, Junior et al. emphasized the importance of lubrication and cutting fluids for tool life³¹, while Nalbant & Yildiz demonstrated that cryogenic cooling decreased cutting forces in AISI 304 milling³². Lastly, Fnides et al. discovered that the most important factor influencing surface roughness in dry milling AISI 1040 was cutting speed³³. According to Lee et al., who milled JIS SS400 steel using canola, soybean, and palm oils, straight oils created the finest finish, palm variants decreased cutting forces, and soluble canola oil produced the highest temperature³⁴.

When machining Al 7050-T7451, Bork et al. discovered that Jatropha oil decreased surface integrity by 5% in comparison to mineral oil³⁵, and Burton et al. observed that emulsified canola oil decreased cutting forces by 22% in the milling of Al 6061 and Steel 1018³⁶. Additionally, Lee et al. demonstrated that refined canola decreased surface integrity loss by 22% and flank wear by 6% on SS400³⁷. Using vegetable-based fluids on Ti-6Al-4 V, Gariani et al. were able to reduce fluid use (42%), cutting force (16.41%), and tool wear (46.77%)³⁸. Werda et al. found that utilizing synthetic ester and fatty alcohol reduced the tip temperature by up to 45% and the cutting force and roughness by 20–40% on X100CrMoV5 steel³⁹. In AISI 420 milling, Uysal et al. reduced tool wear by 16.8% and roughness by 8.8% by combining Nano-MoS₂ with vegetable oil⁴⁰. In comparison to servo oil, Tanneru found that palm kernel oil decreased surface finish degradation by 5% and tool wear by 4%⁴¹. For stainless steels, Senthilkumar et al. demonstrated that cutting speed was dominating in AISI 304⁴², whereas Yusoff et al. observed that feed rate and cutting speed affected wear and roughness in AISI 316L⁴³. Madhavan et al. emphasized the helix angle as the most important tool geometry feature⁴⁴, whereas Karpat et al. showed that speed and feed rate had a significant impact on temperature distribution and integrity in AISI 304⁴⁵. In AISI 304, Rajakumar et al. demonstrated that cryogenic cooling increased tool life and roughness⁴⁶.

Hassan et al. verified speed and feed as important factors using Taguchi methodology⁴⁷. Yusoff et al. demonstrated that cryogenic freezing improved performance with ceramics and coated carbides in AISI 316L⁴⁸, whereas Senthilvelan et al. discovered that coated carbide tools performed better than ceramics in AISI 304⁴⁹. Cutting speed, feed rate, and geometry (edge radius) were shown to be important factors influencing tool wear and roughness in AISI 316L and AISI 304, respectively, when using TiAlN-coated tools by Ahsan et al. and Khodaei et al.^50,51. In AISI 304 machining, Singh et al. demonstrated that Pongamiapinnata oil esters performed better than mineral oil, lowering roughness by 30% and cutting force by 50%⁵².

In AISI 316, neem oil-based lubricants reduced cutting force by 40% and tool wear by 50%⁵³, whereas blends of castor oil reduced roughness by 30% and force by 55% in AISI 304⁵⁴. For AISI 304, waste cooking oil esters reduced force and roughness by 40% and 30%, respectively⁵⁵. In AISI 316, Jatropha oil decreased cutting force by 50% and wear by 60%⁵⁶, while in AISI 304, Mahua oil decreased cutting force by 45% and 40%⁵⁷, Karanja oil decreased cutting force by 50% and 45%⁵⁸, and Castor oil decreased cutting force by 55% and 35%⁵⁹. In AISI 316, Pongamiapinnata oil decreased wear by 40% and force by 50%⁶⁰, but Castor oil reduced wear by 45% and 50%⁶¹. With minimal toxicity and excellent stability, neem oil demonstrated a 40% force and 30% roughness reduction in AISI 304⁶². In AISI 316, synthetic pogomia oil provided biodegradability and thermal stability while reducing force by 50% and wear by 45%⁶³. Jatropha, Mahua, and Karanja oils also showed comparable decreases with additional environmental advantages^64,65,66. Beyond oils, Vardhan et al. confirmed great accuracy by using ANN to predict MRR and surface roughness in P20 steel⁶⁷. MQL experiments revealed that palm oil decreased forces and power consumption over synthetic esters⁶⁸, aloe vera oil increased machinability of M2 steel⁶⁹, mustard oil improved tool life and finish⁷⁰, and coconut oil outperformed wet and dry turning in AISI 1040⁷¹. AISI 316L turning under MQL tested mineral, Simarouba, Pongam, Neem, and other oils in a comparative Taguchi L25 analysis. The most effective oil was neem, which reduced roughness by 21%, wear by 26%, and cutting forces by 20%. Consequently, it was chosen for end milling experiments, where RFR, BPANN, FFANN, DFA, TDOE, and RSM techniques were used for optimization and validation.

Material selection for end milling

The AISI 316L Stainless Steel (SS), as rounded bars, was acquired from Mumbai’s Dhanalakshmi Steel Distributors. The mechanical properties and the chemical makeup of the material are shown in Tables 1 and 2. The AISI 316L microstructure cross-section appears in Fig. 1. For the turning operation, the stainless steel round bar of diameter 50 mm and length 500 mm is used. Further, for the drilling operation, the stainless steel plates of length 80 mm, width 35 mm, and 10 mm thickness were utilized. Finally, for end milling operations, the stainless steel plates of length 80 mm, width 35 mm, and thickness 10 mm are employed as work workpiece.

Table 1 AISI 316L Mechanical Properties.

Table 2 AISI 316L Chemical Compositions.

Fig. 1

The AISI 316L microstructure specimen.

Cutting fluids & MQL

Cutting fluids

Neem oil, mineral oil, Simarouba oil, Pongam oil, and oil–water mixes are among the cutting fluids and lubricants used to mill AISI 316L bars with little to no lubrication. These lubricants’ physicochemical characteristics are shown in Table 3. Using an S-9251 Systonic Viscometer, the lubricants’ viscosity is tested, and a number of results are noted. The dissolved oxygen concentration of neem oil is measured using a diaphragm electrode method. The mass-to-volume relationship is used to determine the lubricants’ densities. Using the Abels flash point tester, the lubricants’ flash points are determined. Neem oil is converted into its ester form, Neem oil methyl ester (MENO), through a transesterification process. In this method, Neem oil is heated and its glycerol content is removed using a 2% KOH catalyst and a 3:1 methanol mixture. Formulated Neem Oil (FNO) is produced by heating MENO once more while adding distilled water to eliminate the alcohol. With a specifically made MQL setup, this prepared Neem oil is used as a cutting fluid and lubricant while drilling and milling AISI 316L.

Table 3 The lubricants’ physico-chemical characteristics.

Minimum quantity lubrication (MQL)

The MQL method was employed to supply straight cutting oil. A mist lubricator with nozzles of 2, 3, and 4 mm in diameter is integrated into the MQL system is utilized. High-pressure oil is provided at different pressures for the machining operation. The MQL system includes an electronic timer, reservoir, filter regulator, pressure switch, variable speed control motor, and a thin pulsed jet nozzle to regulate the frequency of the oil piston pump. The nozzle is positioned on the fixture to ensure that the lubricant directly targets the chip-tool contact during machining. A “specially developed MQL setup” (in-house) delivering lubricant at 10 mL/min, 5 bar pressure, with a 5 mm nozzle standoff is used in this present investigation.

Experimental procedure & set-up for end milling

The experiments on milling are carried out using the AMS Vertical Machining Centre (Model: Spark, Bangalore), utilizing a solid carbide milling cutter (8 mm) (Kennametal — HARVI) as shown in Fig. 2 under MQL conditions. Figure 3 shows an End Milling Machine using an experimental MQL setup. The formulated Neem oil as cutting fluid/lubricant is utilized as a straight cutting oil between the chip and tool interface region through a tailor-made MQL setup under different cutting speeds (m/min), cut depth (mm), and feed velocity (mm/min). The end milling test levels and parameters chosen are recorded in Table 4. The characteristics influencing the average MRR and average surface roughness were identified in order of relevance using the Taguchi method, which was used to implement the test design in this investigation^72,73.

Fig. 2 

Milling tool.

Fig. 3

End Milling Machine experimental MQL setup.

Table 4 Factors and Levels (End milling).

Surface roughness measurement

A surface roughness tester, the Taylor-Hobson Surtronic + 3, was used to quantify the milled surface’s surface roughness (Ra). This device can be powered by either an optional main adaptor or a 9-V alkaline battery. Surface roughness is determined as the average of three measurements taken along the milled surface. The Ra values describe the surface quality of the machined area. The equation used to measure surface roughness is expressed as follows:

$${R}_{a}=\frac{1}{8}\sum_{i=1}^{8}|{x}_{i}|$$

(1)

where x_i is the measured surface roughness.

Material removal rate (MRR) measurement

The material removal rate, or MRR, during the weight method end milling of AISI 316L has been determined using the following methods. Initially, the cutting parameters (depth of cut, feed rate, and spindle speed) were determined by taking into account the material of the workpiece and the tooling. Second, the amount of material to be eliminated was calculated using the following formula.

$$Volume \, of \, material \, to \, be \, removed = \, Width \, of \, cut \, x \, Depth \, of \, cut \, \times \, Length \, of \, cut$$

(2)

Thirdly, material databases or handbooks were used to determine the density of AISI 316L. Finally, the weight of the material that needed to be removed was determined by multiplying its volume by its density.

$$Weight \, of \, material \, to \, be \, removed = \, Volume \, of \, material \, to \, be \, removed \, \times \, Density \, of \, material$$

(3)

Experimental detail

Taguchi’s L₂₇ orthogonal arrays

Taguchi’s L₂₇ orthogonal arrays, which have 27 rows, matching the number of tests with 5 columns at three levels, have been utilized to elaborate the experiment plan. Based on experiments were conducted on Taguchi’s L₂₇array. L₂₇ End milling tests (array row) comprise the experiment plan a shown in the Table 5. The first column was designated as the first input parameter, the second as the second, the third as the third, and the remaining columns as the interaction is as shown in Fig. 4. The output parameter is the response that has to be examined. Material removal rate and surface roughness experimental data were recorded from Table 6. Using the L27 orthogonal array, investigated three primary machining parameters: depth of cut, feed velocity, and cutting speed. Only three of L27’s five columns were devoted to the primary elements; the other two were utilized to illustrate how different parameters interacted with one another. In order to capture both major effects and interactions, the bigger L27 array was used, guaranteeing a more robust and dependable analysis than a smaller array.

Table 5 Taguchi’s Orthogonal array L₂₇ (End Milling).

Fig. 4

Linear graph.

Table 6 Material Removal Rate and Surface Roughness obtained values for different cutting speeds, feed velocity, and depth of cut under formulated Neem oil.

Response surface methodology (RSM)

Analysing the evolution of cutting force during machining is a challenge for the metal cutting industry. Cutting power is the best means of comprehending the machining properties of all metals and alloys. To forecast the cutting force ahead of time, theoretical frameworks that enable forecasting based on operating conditions must be used. Response surface methodology (RSM) is an optimization approach that combines statistical and mathematical methodologies to analyze a problem, build a model, and determine the best cutting condition.20 sets of tests were conducted using a two-level complete factorial with eight factorial points supplemented by an extra six center and six axial points, as seen in Fig. 5, using a face-centered central composite design (CCD). α = 1, three adjustable process parameters (p = 3), and an area of interest coded {− 1, 1} are all part of the experiment; Table 7 shows the values of these variables.

Fig. 5

Representation of a 2³ central composite design.

Table 7 Experimental design matrix.

Desirability functional analysis (DFA)

It is known that the desirability function is a simultaneous optimization method for handling multi-response situations. The goal function in this study first converts the current values into a scale-free value known as desirability. The MINITAB 15 software assesses the composite desirability of a further optimal level of parameters in order to complete specific process output parameters.

Forward artificial neural network (FFANN)

The Feed Forward Artificial Neural Network (FFANN) architecture is shown in Fig. 6. An input layer, hidden layers, and an output layer with unidirectional information flow make up the FFANN. Weighted linkages are used to connect each neuron to the subsequent layer. For precise prediction, it is mostly utilized to map linear and nonlinear correlations in machining parameters.

Fig. 6

Architecture of Feed Forward Artificial Neural Network (FFANN).

Back propagation artificial neural network (BPANN)

Figure 7 shows the Back Propagation Artificial Neural Network (BPANN) Architecture. By using a back propagation technique for learning, the BPANN expands upon the FFANN. It uses a backward pass to reduce errors using gradient descent and a forward pass to produce outputs. It effectively predicts nonlinear processes and optimizes end milling parameters through iterative weight adjustments.

Fig. 7

Architecture of Back Propagation Artificial Neural Network (BPANN).

Random forest regression (RFR) architecture

Figure 8 shows the Random Forest Regression (RFR) Architecture. Multiple regression trees trained on arbitrary subsets of data and features are created using the ensemble technique known as RFR. By averaging the outputs from every tree, final predictions are produced, improving resilience and lowering overfitting. It captures intricate relationships between milling settings quite well.

Fig. 8

Architecture of Random Forest Regression (RFR).

Results and discussions

In many different sectors, end milling is a commonly utilized machining technique for producing AISI 316L stainless steel components. Using a rotary cutting tool with many flutes, material is removed from the workpiece’s surface during this procedure. As Neem oil was discovered to be the best cutting fluid among five cutting fluids in terms of longer tool life, better surface refinement, and lower cutting forces, during turning experiments on AISI 316L, Neem oil is chosen for the modification/formulation. As a result, methyl ester neem oil, or MENO, is created by altering raw neem oil. The AISI 316L was milled using this specially blended oil. Consequently, it is imperative to look into the optimization of the milling output characteristics of AISI 316L under compounded Neem oil utilizing MQL. This section discusses the L₂₇ orthogonal array’s material removal rate and surface roughness when milling AISI 316L under formulated Neem oil.

This study examines how to improve surface roughness and material removal rate when AISI 316L end milling is done using minimal amounts of Neem oil lubricant. By adjusting the cutting rate (m/min), feed velocity (mm/min), and cut depth (mm), Taguchi Design of Experiments (TDOE), Response Surface methodology (RSM), and Analysis of Variance (ANOVA) are used to achieve this. Additionally, Desirability Functional Analysis (DFA), Random Forest Regression (RFR), Back Propagation Artificial Neural Network (BPANN), and Feed Propagation Artificial Neural Network (FFANN) have been used to estimate the material removal rate and surface roughness.

Material removal rate (MRR)

MRR has a direct impact on the effectiveness and caliber of the procedure for machining, making it a crucial consideration in end milling operations. The quantity of material extracted from the workpiece per unit of time is known as the MRR. A speedier machining process, which can boost output and shorten production times, is indicated by a high MRR. On the other hand, a high MRR may also result in shorter tool life and more wear. Moreover, a low MRR may lead to higher manufacturing costs, decreased productivity, and longer machining times. In order to attain the required machining quality and cost-effectiveness, it is crucial to determine the ideal MRR that strikes a compromise between efficiency and tool life. Thus, Figs. 9, 10 and 11 in this part offer a detailed investigation of MRR during end milling of AISI 316L.

Fig. 9

MRR variation with a cutting speed for varying input velocities and a 1 mm cut depth.

Fig. 10

MRR variation with cutting speed for different feed velocities and a 1.5 mm cut depth.

Fig. 11

Change in MRR with cutting speed with varying feed velocity, and a 2 mm cut depth.

Considering the test findings, the maximum material removal rate shall be attained by utilizing MQL in conjunction with the highest cutting speed (150 m/min), feed velocity (250 mm/min), and cut depth (2 mm) while using prepared Neem oil. This could be because of higher feed velocity and cutting speed, which led to enhanced heat dissipation and AISI 316L thermal softening, which in turn increased the rate of material removal, and the same type of results were also seen in^66,67. When assessing the effectiveness of a machining process such as end milling, the Material Removal Rate (MRR) is a critical component. Cutting speed, feed velocity, and depth of cut all have an impact. Increased cutting speed generally leads to a higher MRR because more material can be cut in a given amount of time since the tool material engages with the workpiece more quickly. It can be clearly seen from Figs. 9, 10 and 11 that at a Low cutting speed (50 m/min), the MRR would be lower because the tool is moving more slowly, reducing the amount of material removed per unit time. At the Moderate cutting speed (100 m/min), this would provide a balanced MRR with respect to the tool wear and cutting efficiency. At the High cutting speed (150 m/min), the MRR is expected to increase, but may also result in higher tool wear, especially if the tool material is not optimized for high-speed cutting.

Also from Figs. 9, 10 and 11, a higher feed velocity will directly increase the material removal rate because the tool engages the workpiece with more material being cut per revolution. At Low feed velocity (50 mm/min), the MRR will be lower as less material is fed into the tool per unit time. The MRR will be higher compared to a low feed rate in the case of Moderate feed velocity (100 mm/min). At High feed velocity (150 mm/min), this will likely produce the highest MRR because more material is being fed into the cutting zone. Higher feed velocities and larger depths of cut will consistently increase MRR across all cutting speeds. However, at the high cutting speed (150 m/min), the increase in MRR may begin to plateau if the tool wear or temperature becomes excessive due to a lack of cooling, even with MQL.

Increasing the depth of cut increases the volume of material removed in a single pass, which will directly boost the MRR. From Figs. 9, 10 and 11 clearly show that the values of MRR will be lower because less material is being removed with each pass at a Low depth of cut (1 mm). At a Medium depth of cut (1.5 mm), MRR will be higher due to the increased depth, and the values of MRR are highest, as more material is removed during each tool engagement at a High depth of cut (2 mm). MQL is a lubrication method that uses a minimal amount of lubricant (neem oil in this case), which can help reduce heat and friction in the cutting zone. This is especially important when machining tough materials like AISI 316L. The use of formulated neem oil as a cutting fluid in MQL helps maintain lower cutting temperatures, which can improve tool life and reduce thermal deformation of the workpiece, allowing for higher cutting speeds and feeds without sacrificing performance.

The assessment of the contribution percentage (P%) for the various components considered indicates that cutting speed (m/min) circumstances have the largest contribution for material removal rate, at about 0.717 is as shown in Table 8. As a result, conditions related to cutting speed should be carefully considered when end milling AISI 316L under formulated Neem using MQL. Additionally, after end milling AISI 316L under formulated Neem oil MQL, (B) feed velocity (mm/min) (P = 0.992) and (C) depth of cut (mm) (P = 0.236) also significantly affect the MRR of the steel. As seen in Fig. 12, the maximum material removal rate value was obtained when AISI 316L was end milled using compounded Neem oil with MQL at a speed of 150 m/min, a feed velocity of 250 mm/min, and a depth of cut of 2.0 mm. It shows the primary effects plot for surface roughness. Given processing parameters such as feed velocity (mm/min), depth of cut (mm), and cutting speed (m/min), the second-order response surface (mm³/sec) (Eq. 4) may be defined.

$$\begin{aligned} Material \, Removal \, Rate = & - 0.176{-}0.285637\left( A \right) - 0.126164\left( B \right) \\ & + 14.5989\left( C \right){-}1.45765*10^{ - 4} \left( {A^{2} } \right){-}4.77647*10^{ - 5} \left( {B^{2} } \right) \, \\ & {-}10.3576\left( {C^{2} } \right) + 0.0017\left( {AB} \right) + 0.1435\left( {AC} \right) + 0.0989\left( {BC} \right) \\ \end{aligned}$$

(4)

Table 8 ANOVA for S/N ratios for material removal rate (mm³/sec).

Fig. 12 

MRR Main effects plot (N) (a) S/N ratio.

Table 9 displays the response function material removal rate’s ANOVA results. This analysis is performed with a 95% degree of confidence and a 5% level of significance. Table 9 analysis shows that the generated second-order response function is fairly suitable because the F computed value is greater than the F-table value (F_{0.05, 14, 14} = 144.41). When considering the cut’s depth (1 mm) held constant, the contour and surface plots illustrated in Fig. 13 show the impact of feed velocity and cutting speed on the rate of material removal. Considering the cut’s depth (1 mm) held constant, the contour and surface plots illustrate the impact of feed velocity and cutting speed on the rate of material removal. Figure 13a contour plots demonstrate that the Material Removal Rate (MRR) increases with both cutting speed and feed rate at a fixed depth of cut (1 mm). The contour orientation suggests that feed rate has a greater impact than cutting speed. When both parameters are optimized, the MRR is maximum; when feed and speed are low, the MRR is lowest. Productivity is therefore increased by raising feed rate and cutting speed; nevertheless, optimization is necessary to balance MRR with tool life and surface quality. Figure 13b 3D surface plots demonstrate that MRR rises linearly with both cutting speed and feed rate at a constant depth of cut (1 mm), with feed rate exerting a greater influence. High feed and speed yield the maximum MRR, whereas low feed and speed yield the lowest. This demonstrates the necessity of optimizing parameters to balance productivity, tool life, and surface quality by confirming a direct proportional link.

Table 9 ANOVA for MRR.

Fig. 13

(a) MRR Contour plots (mm³/sec). (b) 3D Surface Plot (mm³/sec).

Surface roughness

The effectiveness of milling is greatly enhanced by a finally milled surface with improved surface roughness, which also increases corrosion resistance, creep life, and fatigue strength. Moreover, surface roughness influences a number of functional characteristics, including wear, heat transmission, light reflection, surface friction caused by contact, lubricant dispersion, coating efficiency, and fatigue life. Consequently, a thorough analysis of the surface roughness following end milling under prepared Neem oil conditions has been covered. Taking into account the experiment results displayed in Figs. 14, 15, and 16, the values with the lowest surface roughness were obtained using a produced Neem oil that employed MQL at a depth of cut of 1 mm, a cutting speed of 100 m/min, and a feed velocity of 200 mm/min. This is because superior cutting conditions have been achieved by lower cutting speeds, faster feed velocities, and reduced cutting depth of cuts. Thermal damage is less likely, and the surface quality is enhanced while cutting at lower speeds since the cutting tool generates less heat. Increasing feed velocities can help improve chip evacuation, smooth down the surface, and reduce cutting forces.

Fig. 14

Surface roughness changes as a function of cutting speed for varying feed velocities and a 1 mm cut depth.

Fig. 15

Surface roughness varies with cutting speed for different feed velocities and a cut depth of 1.5 mm.

Fig. 16

Surface roughness changes as a function of cutting speed for varying feed velocities and a 2 mm depth of cut.

Cutting speed and feed velocity have a significant impact on surface roughness, as shown in Fig. 15. While lower feed rates of 150 and 200 mm/min result in somewhat smoother surfaces, roughness is substantial at low cutting speeds of 50 m/min, especially at a feed velocity of 250 mm/min (~ 5.5 µm). While the roughness at 250 mm/min is still comparatively greater but improved, there is a noticeable decrease in roughness as the cutting speed climbs to 100 m/min, with values falling close to ~ 2 µm for lower feed velocities. The surface roughness values for all feed velocities converge around ~ 2.0 µm at the maximum cutting speed of 150 m/min, indicating that the impact of feed velocity diminishes with increasing cutting speeds. This pattern shows that although feed velocity has a significant impact on surface quality at lower cutting speeds, feed rate has less of an effect on surface roughness at higher cutting speeds because they improve cutting efficiency, encourage AISI 316L thermal softening, and stabilize chip formation.

It can be clearly seen that from Figs. 14, 15 and 16, at Low Cutting Speed (50 m/min), the surface roughness is higher because the cutting tool will engage with the workpiece more slowly. This results in more material being deformed by the tool, leading to increased cutting forces, vibrations, and a rougher surface. However, when using MQL with neem oil, the lubrication reduces friction, which may slightly mitigate surface roughness compared to traditional machining without lubrication. At Medium Cutting Speed (100 m/min), the Surface roughness tends to improve at medium cutting speeds, as the tool’s cutting action becomes more efficient. There is a better balance between material removal and tool wear, allowing for a smoother surface finish. The neem oil with MQL will help maintain a stable cutting action, further improving surface quality. At High Cutting Speed (150 m/min) often results in the best surface finish, as it allows for fast material removal and reduced tool engagement time. However, at very high speeds, the tool may experience more heat and wear. The neem oil used in MQL can provide a cooling effect that minimizes thermal effects and tool wear, which helps in reducing surface roughness.

Also from Fig. 14, 15 and 16, it can be clearly seen that a low feed velocity (50 mm/min) results in smoother surface finishes because the tool moves more slowly, leading to less material deformation per pass. This allows the cutting tool to engage with the workpiece in a more controlled manner, reducing the chances of vibration and tool chatter. The effect of neem oil lubrication is more pronounced, reducing friction and improving the surface finish. A balanced feed rate provides an optimal trade-off between surface finish and material removal rate. At medium feed velocity (100 mm/min), the cutting forces are not too high, and the tool is still able to maintain good contact with the material, resulting in a relatively smooth surface finish. Neem oil with MQL can help reduce friction, providing smoother cuts. At High Feed Velocity (150 mm/min) tends to cause rougher surfaces as the tool moves quickly, leading to greater material deformation and potentially higher vibration forces. The MQL system will still help to reduce friction, but at higher feed rates, the cutting forces may cause vibrations that could lead to higher surface roughness.

The surface roughness values obtained for different depths of cut (1, 1.5, and 2 mm) can be clearly seen from Figs. 14, 15 and 16. A shallow depth of Cut (1 mm) produces a smoother surface because less material is being removed, which reduces the cutting forces and vibration. With MQL and neem oil, the lubrication reduces heat generation and wear, leading to an even smoother finish. Shallow cuts are ideal for achieving high-quality surface finishes. At Medium Depth of Cut (1.5 mm), the material removal rate increases, which can lead to a slightly rougher surface compared to a 1 mm cut. However, the neem oil with MQL will help to maintain lubrication and cooling, potentially keeping surface roughness under control. There may be a slight increase in cutting forces, but the oil should mitigate excessive heat buildup. With deeper cuts, surface roughness typically increases due to the higher cutting forces involved at the Deep Depth of Cut (2 mm). More material is being removed, which can cause increased tool wear and vibrations, leading to a rougher surface. While neem oil in MQL will help with lubrication and cooling, the increased cutting forces at this depth can still result in noticeable roughness.

Neem oil as a lubricant in MQL can improve surface quality by reducing friction. The neem oil lubricates the cutting interface, reducing the friction between the tool and workpiece. This helps in smoother cutting and reduces the chances of scratches, chatter marks, and thermal damage on the workpiece. The minimal application of neem oil through MQL helps to cool the cutting zone, reducing heat generation that could otherwise affect the work piece and tool. This reduces thermal distortion and maintains a more stable cutting environment. Since MQL uses only small quantities of neem oil, it helps in minimizing tool wear by reducing heat and friction, which in turn can contribute to improved surface finishes.

Finally, by minimizing the amount of material removed in a single pass, decreasing the depth of cut can lower the total cutting forces and improve surface roughness. Based on a proportion of the contribution (P%) evaluation of the several components considered, cutting speed (m/min) conditions contribute the most to surface roughness, with an estimated value of 0.781 as shown in Table10. As a result, conditions related to cutting speed should be carefully considered. When end milling AISI 316L under MQL. Moreover, following end milling under formulated Neem oil MQL, (B) feed velocity (mm/min) (P = 0.617) and (C) cut’s depth (mm) (P = 0.351) and their corresponding combinations also significantly affect the surface roughness characteristics of AISI 316L.

Table 10 Surface roughness (micron) ANOVA for S/N ratios.

When AISI 316L was end milled using compounded Neem oil with MQL, the cutting speed (150 m/min), feed velocity (150 mm/min), and cut depth (1 mm) generated the lowest value of surface roughness, according to Fig. 17, which shows the major effects plot for surface roughness. Cutting speed, feed rate, depth of cut, and lubrication method are some of the variables that affect surface roughness when end milling AISI 316L stainless steel using formulated neem oil and Minimum Quantity Lubrication (MQL). By applying a little quantity of lubricant in this example, neem oil to the cutting surface, the MQL approach lowers the total amount of cutting fluid used while maintaining cooling and lubrication at the cutting interface. Processing parameters such as cutting speed (m/min), feed velocity (mm/min), and cut depth (mm) may be used to represent the second-order response surface (Eq. 5), which shows the surface roughness at the micron level.

$$\begin{aligned} Surface \, Roughness \, \left( {microns} \right) = & 8.34650{-}0.0525355\left( A \right){-}0.0786910\left( B \right) + 5.9118\left( C \right) \\ & + 0.000282627\left( {A2} \right) + 0.000280627\left( {B2} \right){-}1.31373\left( {C2} \right) \, \\ & {-}1.14*10 - 4\left( {AB} \right) + 0.0009 \, \left( {AC} \right){-}0.01\left( {BC} \right) \\ \end{aligned}$$

(5)

Fig. 17

Surface Roughness on the surface Plot of main effects (in microns) for S/N ratio.

Table 11 displays the ANOVA result for the response function surface roughness (microns). This analysis is performed with a 95% degree of confidence and a 5% level of significance. Table 11 analysis shows that the constructed second-order response function is fairly acceptable because the F computed value is bigger than the F-table value (F_0.05,9,9 = 16.29).

Table 11 Surface roughness (microns) ANOVA.

The surface and contour plots are shown in Fig. 18, with the cut’s depth (1 mm) held constant, showing the impact of feed velocity and cutting speed on surface roughness. Figure 18a contour plots demonstrate that, at a fixed depth of cut (1 mm), surface roughness increases with increased feed velocity and decreases at higher cutting speeds. High cutting speed and low feed velocity yield the smoothest surfaces, whereas low cutting speed and high feed velocity yield the roughest. Across all cut depths (1 mm, 1.5 mm, and 2 mm), Fig. 18b 3D surface plots demonstrate that surface roughness decreases with increasing cutting speed and increases with higher feed velocity. Because of the increased cutting forces at deeper cuts, variations in roughness are more noticeable.

Fig. 18

(a) Surface roughness Contour plot (microns). (b) Surface roughness 3D Surface Plot (microns).

The goal of end milling AISI 316L was to reach a maximum material removal rate (MRR) of 62.1 mm³/sec and a minimum surface roughness of 2.1 microns, as shown in Fig. 19. After that, a suitable weighting system was utilized to combine the separate desirability functions to generate the composite desirability function. It is determined that under specified Neem oil with minimum quantity lubricating conditions, desirability is obtained with the cutting speed (150 m/min), feed velocity (246 mm/min), and depth of cut (2 mm).

Fig. 19

Composite Desirability for Surface roughness (microns) and Material Removal Rate (mm³/sec).

Chip formation and morphological analysis

Chip formation

A crucial component of end milling that influences the workpiece’s ultimate surface quality and machining performance is chip generation. The type of chips created during AISI 316L end milling, are shown in Fig. 20 depends on a number of cutting parameters, such as feed velocity, depth of cut, and cutting speed. Different sorts of chips are created when the cutting speed is adjusted while maintaining a constant low feed velocity, and depth of cut is shown in Fig. 21. A built-up edge forms at moderate cutting speed because there is less plastic deformation of the material and a lower temperature at which cutting takes place. Surface roughness increases, and the chip’s form becomes uneven as a result. However, at high cutting speeds, the material experiences more plastic deformation and the cutting temperature rises, which causes a thin, continuous chip to develop. This lowers the cutting force and produces a smoother machined surface.

Fig. 20 

Macro images of milled land of AISI 316L under different end milling environments (a) Cutting speed (50 m/min), (b) Cutting speed (100 m/min), (c) Cutting speed (150 m/min) with constant feed velocity (200 mm/min) and cut’s depth (1 mm).

Fig. 21

Chips formed during end milling of AISI 316L under various end milling conditions, (a) Cutting speed (50 m/min), (b) Cutting speed (100 m/min), (c) Cutting Speed (150 m/min) with constant feed velocity (200 mm/min) and cut’s depth (1 mm).

The cutting speed has a direct impact on the development of chips by modifying the temperature at which the material is cut, as well as its plastic deformation. Plastic deformation and continuous chip manufacturing increase in direct proportion to increases in cutting temperature and speed. Since a high cutting speed can also result in increased tool wear and a shorter tool life, finding the optimal cutting speed is crucial to obtaining the necessary machining performance. In conclusion, a critical factor influencing chip formation in AISI 316L end milling is cutting speed. The cutting speed will determine, several varieties of chips can be formed, such as uneven and rough or continuous and smooth. To maximize end milling efficiency and attain the required machining performance, one must comprehend the relationship between chip creation and cutting speed.

Morphological analysis of machined surface

The distribution and arrangement of a material’s constituent grains or crystals is referred to as its microstructure. The mechanical characteristics and overall performance of a material are greatly influenced by its microstructure. Several cutting parameters, such as cutting speed and lubricant type, can affect the microstructure of AISI 316L during end milling. AISI 316L end milling results in a rise in cutting temperature when the cutting speed is increased. Finer grains occur due to the material undergoing more plastic deformation because of the higher temperature. Figure 22 is evidence for this. A more uniform microstructure and better mechanical qualities, including greater strength and hardness, are the outcomes of finer grains. During end milling, the material’s microstructure may also be impacted by the application of minimum quantity lubrication (MQL) using a bio-based prepared Neem oil. A small quantity of lubricant is continually applied to the cutting area using the MQL lubrication method, which lowers wear and friction. Comparing Neem oil MQL to conventional mineral oil-based lubricants, the bio-based formulation offers better lubrication performance and less wear on the cutting tool and workpiece. The combined result of increased pace of cutting and the use of Neem oil MQL results in the creation of finer grains in the material and an improved microstructure, are shown in Fig. 22. The improved microstructure leads to improved mechanical properties and better overall machining performance.

Fig. 22

Microstructure of AISI 316L after end milling with changing cutting speed (a) 50 m/min, (b) 100 m/min, (c) 150 m/min, with constant feed velocity (150 mm/min) and cut’s depth (1 mm).

In conclusion, a range of cutting parameters, such as cutting speed and the kind of cutting oil or lubricant employed, might affect the microstructure of AISI 316L. While using Neem oil, MQL improves lubricating performance and lowers wear, increasing the cutting speed during end milling produces finer grains and an enhanced microstructure. Through meticulous regulation and enhancement of certain cutting parameters, the material’s microstructure can be customized to fulfill particular design specifications and enhance the efficiency of machining.

Validation and prediction using RSM, BPANN, FFANN, and RFR

Tables 12 and 13 demonstrate the validation of RSM, BPANN, FFANN, and RFR with experimental findings for Surface Roughness and MRR. The link among the input variables (cutting speed, feed velocity, and cut’s depth) and the results of the replies is modeled using Response Surface Methodology (RSM), yielding errors of 0.76% and 2.498%, respectively. Using the back-propagation algorithm, a Back Propagation Artificial Neural Network (BPANN) was trained to predict surface roughness and MRR with error rates of 2.267% and 0.57%, respectively. With an error of 1.893% and 2.334%, respectively, the Feed Forward Artificial Neural Network (FF ANN) was also utilized to estimate the MRR and surface roughness during AISI 316L end milling. With an inaccuracy of 0.473% and 0.611%, respectively, the surface roughness and MRR are predicted using the Random Forest Regression (RFR) machine learning technique.

Table 12 Experimental, RSM, BPANN, FFANN, and RFR values of Surface roughness (microns) during end milling of AISI 316L under formulated Neem oil with MQL.

Table 13 Experimental, RSM, BPANN, FFANN, and RFR values of Material Removal Rate (mm³/sec) during end milling of AISI 316L under formulated Neem oil with MQL.

According to the comparison from Fig. 23, all models forecast surface roughness trends that are reasonably similar to the experimental values. While RSM and RFR exhibit somewhat more variance in certain trials, BPANN and FFANN yield the most accurate findings. This demonstrates how much more accurate ANN-based methods are at predicting surface roughness.

Fig. 23

Comparison of Experimental Surface Roughness (microns) with RSM, BPANN, FFANN, and RFR prediction.

The comparison demonstrates in Fig. 24 that the experimental trend for MRR is closely followed by all models (RSM, BPANN, FFANN, and RFR). RSM and RFR exhibit somewhat greater deviations, whereas ANN-based models (BPANN, FFANN) provide the most accurate forecasts, particularly at higher levels. This suggests that nonlinear machining behavior is better captured by ANN techniques.

Fig. 24

Comparison of Experimental Material Removal Rate (mm³/sec) with RSM, BPANN, FFANN, and RFR prediction.

In general, Figs. 23, 24 and 25 show the accuracy of all four approaches in forecasting the rate of material removal and surface roughness during AISI 316L end milling is good to acceptable. But it was discovered that Random Forest Regression had the lowest error, making it the most appropriate technique. It was found that using prepared Neem oil MQL worked well while machining end milling AISI 316L. The output parameters of MRR and Surface Roughness were greatly changed by changing the input parameters, which included cutting force, feed velocity, and depth of cut. RSM, BPANN, FFANN, and Random Forest Regression were all employed in the optimization process; of these, Random Forest Regression proved to be most appropriate because of its excellent accuracy in output response prediction. The claim of Neem oil MQL provided improved lubrication performance, reducing friction and wear during the end milling process. This, in turn, led to better surface finish, increased material removal rate, and improved overall machining performance. The microstructure of the material was also proven to be influenced by the machining parameters, with a developed cutting speed producing finer-grained and better-structured material.

Fig. 25

Percentage Errors between the material removal rate and surface roughness.

Figure 25 Percentage Errors between the material removal rate and surface roughness values that were expected while AISI 316L was being end milled using formulated Neem oil with MQL. Furthermore, the claim of Neem oil MQL was discovered to affect chip formation, reducing the amount of built-up edge and improving the quality of the chips produced during end milling. The combination of improved lubrication performance and microstructure alterations led to improved machining performance and better overall results. It is possible to efficiently get the ideal process parameters of entry for a higher MRR and a reduced surface roughness by using the designed L₂₇ Orthogonal array. When milling AISI 316L, the RSM, BPANN, FFANN, and RFR models have accurately predicted the material removal rate and surface roughness.

Conclusions

AISI 316L stainless steel was subjected to end milling trials in this study with different cutting settings using neem oil as a lubricant. Using an L27 orthogonal array, the tests were planned. The RFR, FFANN, BPANN, and RSM models were used to accurately estimate the surface roughness and material removal rate (MRR). The outcomes of the analysis were as follows.

• MRR is a critical factor in end milling operations, directly influencing machining efficiency. High MRR leads to faster production but may cause increased tool wear, while low MRR can reduce productivity and increase manufacturing costs.

• Higher cutting speeds (150 m/min) resulted in increased MRR due to better heat dissipation and thermal softening of AISI 316L, although this could lead to higher tool wear. Conversely, low cutting speeds (50 m/min) yielded lower MRR.

• Higher feed velocities (150 mm/min) consistently enhanced MRR by engaging the workpiece more effectively, whereas lower feed velocities (50 mm/min) resulted in reduced material removal rates.

• A greater depth of cut (2 mm) significantly increased MRR by removing more material per pass. Shallow cuts (1 mm) resulted in lower MRR due to less material being removed.

• Minimum Quantity Lubrication (MQL) using prepared Neem oil effectively reduced heat and friction, improving tool life and enabling higher cutting speeds and feed rates without compromising performance.

• Cutting speed (150 m/min), feed velocity (250 mm/min), and cut depth (2 mm) were identified as the optimal conditions for achieving the highest MRR in the end milling of AISI 316L.

• The lowest surface roughness values were achieved at medium cutting speeds (100 m/min), moderate feed velocities (200 mm/min), and low depth of cut (1 mm). The use of Neem oil in MQL helped reduce thermal damage and improve surface quality.

• Cutting speed was the most significant factor influencing surface roughness, with higher cutting speeds typically leading to better surface finishes when combined with MQL and Neem oil lubrication.

• The cutting speed influenced chip formation, with higher speeds producing continuous, thinner chips, reducing cutting forces, and improving surface quality. Lower speeds led to uneven chips, resulting in rougher surfaces.

• RSM, BPANN, FFANN, and RFR were employed to optimize cutting parameters. Random Forest Regression (RFR) provided the most accurate predictions for both MRR and surface roughness, highlighting its effectiveness in process optimization.

In summary, the findings of this study suggest that Neem oil can function as an efficient lubricant/cutting oil cutting performance under MQL conditions. The developed L₂₇ orthogonal arrays, RSM, ANOVA, DFA, BPANN, FFANN, and RFR can be effectively utilized to AISI 316L end milled to determine the best process input parameters and to forecast MRR and surface roughness.