RESEARCH OF MULTI-RESPONSE OPTIMIZATION OF MILLING PROCESS OF HARDENED S50C STEEL USING MINIMUM QUANTITY LUBRICATION OF VIETNAMESE PEANUT OIL

This study aims to build a regression model when surveying the milling process on S50C steel using Minimum Quantity Lubrication (MQL) of Vietnamese peanut oil-based on Response Surface Methodology. The paper analyses and evaluates the effect of cutting parameters, flow rates, and pressures in minimum quantity lubrication system on cutting force and surface roughness in the milling process of S50C carbon steel materials after heat treatment (reaching a hardness of 52 HRC). The Taguchi method, one of the most effective experimental planning methods nowadays, is used in this study. The statistical analysis software, namely Minitab 19, is utilized to build a regression model between parameters of the cutting process, flow rates and pressures of the minimum quantity lubrication system and the cutting force, surface roughness of the part when machining on a 5-axis CNC milling machine. Thereby analyzing and predicting the effect of cutting parameters and minimum quantity lubrication conditions on the surface roughness and cutting force during machining to determine the influence level them. In this work, the regression models of Ra and F were achieved by using the optimizer tool in Minitab 19. Moreover, the multi-response optimization problem was solved. The optimum cutting parameters and lubricating conditions are as follows: Cutting velocity Vc = 190.909 m/min, feed rate fz = 0.02 mm/tooth, axial depth of cut ap = 0.1 and nozzle pressure P = 5.596 MPa, flow rate Q = 108.887 ml/h. The output parameters obtained from the above parameters are Ra = 0.0586 μm and F = 162.035 N, respectively. This result not only provides the foundation for future research but also contributes reference data for the machining process.


Introduction
S50C steel material is high strength steel, has medium carbon content, has strong oxidation resistance, high rust and heat resistance. Another advantage of this is good polishing. For these reasons, S50C steel is widely used in the mould industry for the manufacture of injection moulds, plastic moulds. Moreover, it also used in shipbuilding, components for factory structure, and other mechanical parts, etc. There are many different methods to generate products from S50C steel materials, such as pressure forming process, casting, metal cutting. Among these methods, metal machining accounts for a relatively high proportion in the automotive manufacturing industry and machine components.
It is necessary for manufacturing enterprises to approach comprehensive researches on issues concerning economic and production efficiency in order to obtain solutions to these problems. For the reason of meeting the rocket development of the industry and the demand of highly competitive market nowadays, it is necessary to apply a reasonable plan in the machining process.
Hardened steel is generally a difficult-to-cut material, but the technology of machining this material with a tool that has cutting edges, typically hard milling, has attracted considerable attention.

Engineering
In machining and manufacturing of cores and cavities of moulds, hard milling has emerged as an alternative machining method to Electrical Discharge Machining (EDM), which has a high cost and a long period of time in Machining [1,2]. To mill hard materials, it is imperative that we have the right equipment as well as appropriate solutions, and the need for minimum quantity lubrication technology is leading to great concerns now [3,4]. In this paper, the author emphasizes studying the simultaneous effects of cutting parameters, flow rates, and injection pressure of the minimum quantity lubrication system on surface roughness and cutting force when milling S50C steel materials after heat treatment.
The purpose of this study is to build a mathematical regression model between the input parameters: cutting speed (V c ), tooth infeed amount ( f z ), cutting depth in the axial direction (a p ), pressure (P), and flow rate (Q) with surface roughness and cutting force in output when milling S50C steel after heat treatment.
The process of cutting hard materials with the indexable cutting tool experienced continuous changes in the cutting force in the whole cutting process. There are many different factors inside and outside the machining system that directly or indirectly affect the cutting force in the machining process, and some typical parameters are synthesized by the fishbone diagram in Fig. 1 [5,6].
Additionally, the surface roughness of the part is also one of the pivotal criteria for product evaluation. Recent studies show that the use of minimum quantity lubrication in milling has positive results in product quality and environment [7][8][9]. Over the last couple of decades, many different optimization methods were developed and presented in order to optimize surface roughness in machining, including milling, turning, and grinding. A growing market has made it necessary for manufacturers to improve product quality as well as reduce cost. Hence, multi-objective Optimization became more popular in recent years. Optimization of the milling process, especially when investigating hard materials or modern lubrication methods, is receiving a lot of attention from researchers and manufacturing companies. Optimizer approaches vary depending on the researcher's expectation of the targeted model. It has been proposed that the particle swarm optimization algorithm (PSO) be used to extract the optimal parameters in view of the boundary conditions, which can be determined through the empirical relationship of the factors that affect machining. The authors in reference [10] have used that method to show an optimization study on optimum determination of the cutting parameters and cutter helix angle when end milling of Al6061. Research on the Optimization of MQL-enabled milling processes is also attracting a great deal of interest. Mozammel Mia has been successful in Mathematical Modelling and Optimization of MQL assisted end milling characteristics based on RSM and Taguchi method [11]. In [12], the study carried out research applying RSM in multiple-response Engineering Optimization in end milling of S50C medium carbon steel assisted MQL, concentrating on minimizing residual stress, cutting force and surface roughness. There is a correlation between the cutting force in milling and the surface roughness of the workpiece, so it is possible to rely on the tendency of the cutting force to predict the changing pattern of the surface quality of the workpiece [13]. Thus both surface roughness and shear force must be monitored during the machining process, thereby aiming to solve the problem of online monitoring and multi-target Optimization of quality and technical indicators to help increase economic efficiency and product quality when machining [14].

Materials and methods 1. Surveying condition
The 5 axis machining centre CNC (DMU50) with the control system of Siemens 840D (Fig. 2  Minimum Quantity Lubrication System (MQL): minimum lubrication system for metalworking of TOPSET, Beijing, China, is shown in Fig. 3. Specifications are as follows: Nozzle size: 6 mm, maximum number of nozzles: 2 nozzles. The lubricant of the minimum lubrication system in this study was peanut oil.
Cutting Tool: the experiment using EGO ® Indexable End Mills with a WIDA brand cutting insert, which under the code APMT1135PDR-SPTIEH from India. Cutting diameter (d) = 16 mm, body diameter (D) = 17 mm. The number of insert positions: 2, length: 150 mm. WIDA carbide inserts coated with TiAlN, tip radius R = 0.8 mm. Fig. 4 depicts the cutting tool used in this study.

Fig. 4. Cutting tool
Cutting force measurement system: the force sensor of Swiss company Kistler is used to measure the cutting force of three components F x , F y , F z . With the force measurement range from -3 KN to 3 KN, the system will receive data type 3160-B-042, then DynoWare software is used to collect and analyze the force data. The force measurement system is installed following the description in Fig. 5. Engineering Surface roughness measurement system: the surface roughness R a was measured by MITUTOYO-Surftest SJ-210 Portable Surface Roughness Tester. SurfTest SJ USB Communication Tool Ver5.007 software allows displaying and storing measurement parameters R a according to ISO 1997 standard. Each experiment was measured three times. The average value of 3 times was used to analyze and evaluate the experimental results. Details of the surface roughness measurement system are displayed in Fig. 6. Workpiece: experiments are carried out on JIS S50C carbon steel after heat treatment (shown in Fig. 7), which has 52 HRC hardness. Size of test sample L×W×H = 70×30×15 (mm). The chemical composition of S50C steel in Table 1 and the specification of S50C carbon steel are shown in Table 2. Table 1 The chemical composition of S50C steel  With the above properties and the hardness that S50C steel achieves after heat treatment, S50C steel creates difficulties in machining, especially in methods that use a cutting tool, milling for instance.

Experimental method
Through the research model, the study conducted experiments with V c , f z , a p , P and Q. Applying the empirical method Taguchi L27 orthogonal with three different levels to experimentally analyze and predict cutting force and surface roughness when milling. Based on the Taiwanese EGO ® carbide cutting tool recommendation of the cutting tool manufacturer, the cutting parameters for testing S50C steel material after heat are within the following limits: -the cutting velocity V c on the milling machine in the range: 120÷300 m/min; -the cutting depth in axial direction a p : 0.1÷0.9 mm; -the advance f z in the range: 0.02÷0.1 mm/tooth. Based on the facilities of the workshop and capabilities of the MQL system, the limits for the lubrication system parameters are: -nozzle pressure P: 2÷6 MPa; -flow rate Q: 50÷150 ml/h. According to Taguchi orthogonal experimental planning theory, an experiment with three levels is selected and determined in Table 3. In experimental research, with five input parameters, each parameter includes 3 different levels, the most suitable experimental matrix is (L27-3 5 ), including 27 selected experiments, which are shown in Table 4. Since then, we are conducting experimental research on the effects of 5 parameters, including cutting speed, feed rate, axial cutting depth of machining process and pressure, the flow rate of minimum quantity lubrication system on cutting force and surface roughness in the milling process.

1. Analysis of surface roughness
The results of the analysis of variance for surface roughness when machining S50C carbon steel after heat with indexable milling cutters are listed in Table 5. The analysis results in these tables show that: feed rate ( f z ) has the greatest influence on the surface roughness of parts after machining, with 20.56 %, the effect of cutting speed (V c ) accounts for 4.10 %, depth of axial cutting (a p ) comprise 15.14 %, the pressure of minimum quantity lubrication system (P) takes up 6.03 %. Besides, the effect of the square of the cutting force (V c ·V c ) makes up 16.68 % and the square of the feed rate ( f z ·f z ) registries 12.14 %, while other parameters affect less than 6 % are calculated in the ANOVA analysis table (  The regression equation, which expresses the effect of cutting parameters and minimum lubrication condition on the surface roughness, is established. It illustrates the influence of individual parameters and the mutual interaction among inputs to surface roughness, and the summary is evaluated in ANOVA analysis Table 5. The comparison of the predicted results and the measured results of the surface roughness of the part after machining is described in Fig. 8. Run no.

Measured Ra Predicted Ra
Original Research Article: full paper (2021), «EUREKA: Physics and Engineering» Number 6 Engineering Through the description model, it shows that the predicted results are very close to the measured results. The value of R 2 of the regression equation of surface roughness reaches 93.12 %. Therefore, this mathematical regression model is the most suitable regression model with five input parameters (cutting speed, axial depth of cut, feed rate, flow rate and pressure) and the surface roughness, which is known as an output parameter:

2. Analysis of cutting force results
Using the ANOVA method to analyze the influence of parameters through Table 6 shows that the axial depth (a p ) has the greatest influence on cutting force, at 62.61 %, the feed rate ( f z ) registries 16,69 % and cutting speed (V c ) takes up 4.61 % , while other parameters are shown in ANOVA analysis table. The regression equation, which expresses the effect of cutting parameters and minimum lubrication condition on the surface roughness, is constructed. It depicts the influence of individual Engineering parameters and the mutual interaction among inputs to cutting force, and the summary is evaluated in ANOVA analysis Table 6. The comparison of the predicted results and the measured results of the cutting force is described in Fig. 9.

Fig. 9. Measured vs Predicted value of cutting force F
Through the description model, it shows that the predicted results are very close to the measured results. The value of R 2 of the regression equation of surface roughness reaches 97.63 %. Thus, this mathematical regression model is the most suitable regression model with five input parameters (cutting speed, axial depth of cut, feed rate, flow rate and pressure) and the cutting force, which regarded as another output parameter of the model: (2)

Multi-respond Optimization of surface roughness and cutting force 1. Surface roughness (R a )
Taguchi analysis method of Minitab software is used to find out the influence of pressure (P), flow rate (Q), cutting speed (V c ), feed rate( f z ) and axial cutting depth (a p ) to surface roughness (R a ) ( Table 7). As a result, we have a chart in Fig. 10 to evaluate the influence of cutting parameters and the lubrication conditions on surface roughness. The graph shows that the surface roughness tended to decrease slightly from 0.14 µm to less than 0.13 µm while puts the pressure (P) from level 1 to level 2. And a significant growth is seen in the surface roughness when P moves up to level 3, increasing to more than 0.17 µm. When the flow rate (Q) underwent an increase from level 1 (50 ml/h) to level 2 (100 ml/h), the surface roughness decreases considerably from 0.16 µm to 0.14 µm. However, after that, this value tends to climb slightly to more than 0.14 µm at level 3 of the flow rate. Cutting velocity (V c ) parameters have a similar pattern to pressure (P) and flow rate (Q), but V c shows a more pronounced effect. Cutting velocity grew from level 1 (120 m/min) to level 2 (200 m/min) made the surface roughness has had a significant improvement. The surface roughness value decreased noticeably from nearly 0.18 µm to approximately 0.11 µm. Surface roughness is influenced most by feed rate ( f z ). Specifically, the graph shows that the surface roughness increased sharply when set the feed rate from level 1 to level 2, climbing from 0.07 µm to 0.18 µm. It hits the peak at that value and decreases slightly to 0.16 µm when adjusting the feed rate to level 3. Ranked third among the most influential parameter for surface roughness is the axial depth of cut (a r ). The statistics illustrate that if the depth of cut increases gradually from level 1 (0.1mm) to level 2 (0.5 mm) and level 3 (0.9 mm), the surface roughness witnessed a rapid increase to 0.15 µm and 0.17 µm, respectively.

2. Cutting force (F)
Similarly, we analyze to find out the influence of pressure (P), flow rate (Q), cutting velocity (V c ), feed rate ( f z ) and cutting depth in the axial direction (a p ) on the cutting force during the milling process (Table 8). Thus, we got a chart that evaluates the influence of cutting parameters and minimum lubricating parameters on the cutting force when machining (Fig. 11). The biggest influence on the cutting force in the surveyed parameters belongs to axial cutting depth (a p ). The cutting force increases rapidly from 280 N at level 1 (0.1 mm) up to approximately 800 N at level 3 (0.9 mm). Followed close to that, the feed rate ( f z ) and the cutting velocity (V c ) ranked 2 nd and 3 rd , respectively. Both these two factors boost up the cutting force F. Noticeable growth in shear force (F), about 300 N, is reported in Fig. 11. That increase makes the cutting force (F) reaches more than 600 N at the third level (0.1 mm/tooth) of the feed rate ( f z ), from just over 300 N at level 1 (0.02 mm/tooth). It should be noted that cutting speed (V c ) has the same tendency to affect cutting force (F) with feed rate ( f z ), starting with a value of about 450 N at the first level, the cutting force F has increased to 500 N at the second level (210 m/min) and continues to increase when adjusting the cutting speed (V c ) to the 3rd level (300 m/min) reaches nearly 600 N.
While the influence of the cutting parameters on the cutting force during machining shows a strong and pronounced impact, the effects of two parameters of the MQL system, including nozzle pressure (P) and oil flow rate (Q), on cutting force (F) is insignificant. Cutting force (F) changes through the investigated levels of pressure (P) and flow rate (Q) is not much. Over three levels of flow rate (Q), the cutting force (F) increases moderately, and the opposite is true when observing the cutting force at various pressure (P) levels. However, a fluctuation around 500 N was experienced in the value of F in those cases.

Multiple response optimization
In this study, Minitab 19 is employed here to solve multiple-object optimization problems by using the response surface optimization tool. The results of the analysis are shown in Tables 9-12 and Fig. 12. Statistical data from Tables 9-12 show that the surface roughness and cutting force are minimized at the value of R a = 0.0586 µm and F = 162.035 N. To achieve mentioned optimum results, the values of the parameters in the milling process are the cutting speed V c = 190.909 m/min, feed rate f z = 0.1 mm/tooth, depth of cut in the axial direction a p = 0.1 mm. Along with that are the parameters of the lubrication system, including nozzle pressure (P) and flow rate (Q), which are 5.595 MPa and 108.887 ml/h, respectively.

Discussion
The analysis results for surface roughness showed that the influence of cutting parameters, including V c , f z and a p , on surface roughness is substantial, while the settings for the minimum lubrication system make a trivial impact (Table 7, Fig. 10). During machining, surface roughness is characterized by scratches that cutting tools leave on the surface of the part after machining. Obviously, the minimum quantity lubrication system is not directly involved in that process, so it affects restrictedly the surface roughness.
The analysis results for the cutting force illustrated that this output parameter was influenced considerably by the cutting parameters, especially the axial depth of the cut -a p (Fig. 11).

Engineering
In contrast, the influence of the minimum quantity lubrication system parameters was insignificant ( Table 8). The cutting force during machining is known as the measured value from the interaction between the cutting tool and the workpiece under a certain situation (cutting condition). In this case, the minimum lubrication system acted as a supporter in the machining to make it worked more efficiently. Thus their effect may not be apparent.
Based on the multiple response optimization results, with the criteria of minimizing both surface roughness and cutting force, it is statistically reported that the factors that were considered to have the greatest influence on surface roughness and cutting force, which were f z and a p in respective order, were predicted to set in the lowest level (Fig. 12). This demonstrates the consistency of the study.
In this study, the effects of parameters of the minimum lubrication system on surface roughness and cutting force are not clear (Fig. 10, 11). This research stops at the evaluation of the milling process of S50C steel after heat treatment with the aid of MQL. And it also wants to find out how the MQL system impacts the machining process, while it already has various recognized advantages such as minimizing cooling fluids waste and impact positively on the environment. Therefore, the study focuses on evaluating the influence of input parameters on output parameters and predicting multi-object optimization settings for output parameters.
It is also important to notice that the investigation range of the MQL system parameters is limited by the workshop facilities and experimental equipment. Therefore, it is possible that the study may not touch the values that cause a huge change in results yet. This study is limited to an experimental study investigating five input parameters, including three cutting parameters (V c , f z , a p ) and two parameters of the MQL system (P and Q). Other important machining parameters such as system vibration, tool wear, material removal rate, etc. have not been included to evaluate cutting force and surface roughness in this study. In the future, comprehensive studies on input and output parameters, in order to optimize productivity, tool life, etc in the hard milling process, will be the directions for further development of this research.

Conclusions
The paper has conducted experiments to survey and analyze the effects of cutting parameters as well as parameters of minimum lubrication conditions (V c , f z , a p and Q, P) when milling S50C hardened steel after heat treatment. The research applied Taguchi experimental planning method to design and evaluate experiments. Besides, the software called Minitab 19 was used as a tool to build experimental regression mathematical models between input parameters (consist of cutting parameters and MQL conditions) and output parameters, which are the product's surface roughness and cutting force during machining, depicting in equations (1) and (2) in respective order.
The results show that the surface roughness is majorly affected by feed rate f z (20.56 %), cutting speed V c (4.10 %), depth of cut axial direction (15.14 %), minimum quantity lubrication system pressure P (6.03 %), injection oil flow rate Q (1.63 %) and square of cutting speed V c ·V c (16.68 %), squared of feed rate f z ·f z (12.14 %).
Cutting force during machining (F) mainly corresponds to the influence levels of the depth of cut in the axial direction (62.61 %), feed rate f z (16.69 %), cutting velocity V c (4.61 %), oil flow rate Q (1.12 %), nozzle pressure P (0.42 %) and interaction between feed rate and axial cutting depth a p (7.70 %).
The combining milling parameters and lubricant condition as following: the cutting velocity V c of 190.909 m/min, feed rate f z of 0.1 mm/tooth, axial depth of cut of 0.1 mm and oil flow rate, air pressure of 108.887 ml/h, 5.595 MPa, respectively. Corresponding to the surface roughness R a and cutting force F of 0.0586 µm, 163.035 N.