Next Article in Journal
The Effect of Incorporating Silica Stone Waste on the Mechanical Properties of Sustainable Concretes
Previous Article in Journal
Fast Response Isopropanol Sensing Properties with Sintered BiFeO3 Nanocrystals
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

MQL Strategies Applied in Ti-6Al-4V Alloy Milling—Comparative Analysis between Experimental Design and Artificial Neural Networks

1
Faculty of Mechatronic Technology, National Service for Industrial Training (SENAI-SP), São Caetano do Sul, SP 09572-300, Brazil
2
Department of Mechatronics and Mechanical Systems Engineering, Polytechnic School of Engineering of the University of Sao Paulo (USP), São Paulo, SP 05508-900, Brazil
3
Department of Mechanical Engineering, University Center of the Mauá Institute of Technology (IMT), São Caetano do Sul, SP 09580-900, Brazil
4
Technology Faculty of Mauá, The Paula Souza State Center for Technological Education (CEETEPS), São Paulo, SP 09390-120, Brazil
5
Department of Mechanical Engineering, University Center of Educational Foundation of Ignatius (FEI), São Bernardo do Campo, SP 09850-901, Brazil
6
Department of Mechanical Engineering, University of Taubaté (UNITAU), Taubaté, SP 12020-270, Brazil
7
Institute for Innovation in Advanced Manufacturing and Microfabrication, National Service for Industrial Training (SENAI-SP), São Paulo, SP 04757-000, Brazil
*
Authors to whom correspondence should be addressed.
Materials 2020, 13(17), 3828; https://doi.org/10.3390/ma13173828
Submission received: 2 August 2020 / Revised: 20 August 2020 / Accepted: 26 August 2020 / Published: 30 August 2020
(This article belongs to the Section Manufacturing Processes and Systems)

Abstract

:
This paper presents a study of the Ti-6Al-4V alloy milling under different lubrication conditions, using the minimum quantity lubrication approach. The chosen material is widely used in the industry due to its properties, although they present difficulties in terms of their machinability. A minimum quantity lubrication (MQL) prototype valve was built for this purpose, and machining followed a previously defined experimental design with three lubrication strategies. Speed, feed rate, and the depth of cut were considered as independent variables. As design-dependent variables, cutting forces, torque, and roughness were considered. The desirability optimization function was used in order to obtain the best input data indications, in order to minimize cutting and roughness efforts. Supervised artificial neural networks of the multilayer perceptron type were created and tested, and their responses were compared statistically to the results of the factorial design. It was noted that the variables that most influenced the machining-dependent variables were the feed rate and the depth of cut. A lower roughness value was achieved with MQL only with the use of cutting fluid with graphite. Statistical analysis demonstrated that artificial neural network and the experimental design predict similar results.

Graphical Abstract

1. Introduction

Machining processes, such as turning, boring, drilling, and milling, are some of the most important techniques used in industries [1]. In these processes, the material is separated from the original body by the chips, creating a new shape, and it is estimated that it contributes to approximately 5% of the GDP in developed countries [2]. This removal of wanted parts leads to plastic deformations of work specimens, and most of the fed energy is converted into heat [3].
Considering the importance of these processes in the industry, mainly in aerospace, which is highly specialized, it is necessary to keep resources and developments to meet the demand when it arises [4]. Most materials used in the aerospace industry are hard to machine [5]. The Ti-6Al-4V alloy with an α-β microstructure is estimated to account for 50% of the global titanium metal production, and 80% of this corresponds to the aerospace and medical industries [6,7]. According to [8], 14% of the aerostructure of the Airbus A350-900 XWB [2], 15% of the Boeing 787 [3], and 25% of the GE CF6 aeroengine is made from titanium alloys. In addition to the aerospace industry, titanium is also used in the marine, biomedicine, energy, automobile, and chemical industries [9,10]. The main properties of titanium are a high hot strength and hardness, and a superior corrosion resistance and fracture toughness when compared with other materials. It has the highest strength-to-weight ratio amongst all structural materials, excellent creep and biocompatibility, a low elastic modulus, and chemical inertness at room temperature [11]. As a result of these properties, titanium is considered a hard-to-cut material [12], and the main problems include the fact that the high chemical reactivity makes chips easily adhere to the tool-cutting edge, and the low thermal conductivity (1/6 of that for steel) increases the temperature at the tool’s cutting edge (which can easily reach beyond 1000 °C) [13,14,15,16]. A low modulus of elasticity and an extreme strength at high temperature generate long ductile chips, a relatively large contact length between the chip and the cutting tool, and a high compressive stress, leading to a poor tool life, or to higher cutting forces [17].
Since heat generated during machining is a major problem, it is important to choose an adequate lubrication/cooling condition, to improve the machinability of a difficult-to-cut material [12,18]. The minimum quantity lubrication (MQL) is an attractive alternative solution for floods, which have occurred for decades and bring health risks to workers [17,19], and for dry machining, which can accelerate tool wear and the degradation of piece surface integrity [11]. This technique has also been used more often, due to pressure from governments, law enforcement, and environmental agencies that have compelled manufacturers to adopt it [20]. It uses a small amount of cutting fluid in the form of an oil mist that is directly sprayed to the contact zone, helping to reduce friction and cutting forces, and decreasing the surface’s crumple zone by a large amount (~50%) when compared with dry cutting [21]. Consequently, MQL reduces tool wear [22,23]. This technique combines the lubrication from the cutting fluid and the cooling from the air pressure, sometimes reducing the thermal shock caused by flood lubrication [24], and can be used in drilling, milling, turning, and others. Studies conducted by [25] also showed a minimization in vibrations along with radial force during turning.
Artificial neural networks (ANNs) model the complex nonlinear relationships between input and output parameters by observing datasets and identifying patterns, without the need to write explicit programs [4]. An ANN is inspired by the way biological nerves, such as the brain, work to solve problems, and the first artificial neuron was produced in 1943 by McCulloch and the logician Walter Pits [26,27]. The ANN and the genetic algorithm (GA) are important alternatives to be used in machining processes, due to their high complexity in optimizing cutting parameters. They have become very popular in recent years due to their capability of learning nonlinear behavior [28], with many studies conducted. The authors in [27] studied ANNs to predict the material removal rate and surface roughness in the CNC milling of P20 steel and the results indicating a successful application. The authors in [9] used a backpropagation neural network (BPNN), and optimized the overall cutting performance during the high-speed turning of the Ti-6Al-4V alloy, and the results achieved a balance among all studied responses. The authors in [18] investigated the characteristics of different nano-cutting fluids, molybdenum disulfide, and graphite under MQL conditions, using Box–Cox transformation, normal probability plots, and analysis of variance (ANOVA) tests. The best choice for the application was chosen, and the machining characteristics of the Inconel-800 alloy were improved. The authors in [17] studied turning tests with a combination of cooling techniques on titanium (Grade 2). The surface quality measurements, force values, and tool wear were investigated using a combination of ANOVA, and the correlations were established with a success rate of 90%. The authors in [28] used regression analysis and an ANN to predict the surface roughness during the hard turning of AISI 52100, and found superior results in the ANN application. The authors in [4] used the friction force computed for 648 experimental trials to develop and optimize ANN architecture prediction during turning with different materials and cutting parameters, and obtained a 12.5% difference between the estimated and measured values. The results obtained in [24] were even better, with variations from 1% to 4% during a comparative experimental study in three different machining environments (dry air cooling, flood cooling, and minimum quantity cutting fluid). The authors in [29] studied the dry machining of hardened steel En31 to identify the influence of input parameters (speed, feed rate, and the depth of cut) on the surface roughness, and it was possible to analyze the process sensitivity. The authors in [30] used an ANN to predict the cutting force by monitoring the spindle and the individual drives, obtaining success in application during 2.5D milling in E355 steel. The authors in [31] also used an ANN and a GA to predict the minimum surface roughness and chose the best cutting parameters to achieve it. The authors in [32] described numerous papers in which an ANN was developed using this modern technique for many applications besides machining, showing its versatility.
Over time, statistical tools have been applied and have helped to assist in the process of analyzing problems and making decisions. The design of experiments (DOE) has often been used to design a specific experiment or to define data in a methodology. The collection and organization of data for the experiment should ensure that the process has the best possible yield and that there is safety or a minimal difference in the empirical results related to the collected data.
As demonstrated by a growing number of articles published in the last decade, many researchers in the manufacturing field explore ways of applying artificial neural networks to control or to perform an estimation of a product’s critical quality, and to optimize production processes [33]. In this respect, many authors use DOE techniques to implement and optimize ANN parameters. According to authors in [34] the combination of multilayer perceptron with DOE has proven to be an appropriate tool in modeling and problem analysis. Chromium layer thickness predictions on hard chrome plating processes´ results suggest that DOE may be successfully used for the optimization of ANNs´ backpropagation parameters. In another paper on turning, DOE was employed to determine factor levels that benefit network forecasting skills, concluding that the DOE methodology constitutes a better approach to the design of RBF networks for roughness prediction than most trial and error common approaches [35]. Authors in [36] comment that an efficient methodology is needed to obtain optimal values for various parameters in artificial neural networks. In their paper, authors used the Gray–Taguchi method to determine the optimal value of various ANN model input parameters trained by different algorithms in a 2.5D milling process modeling.
The results showed that the combination with RNA presents a better performance. In the search for better prediction values, some studies compare statistical methodologies to ANNs. Authors in [37] studied the AISI316L steel dry turning surface roughness prediction, by comparing ANNs to multiple regression methods. The results found by the methods indicate a better artificial neural networks accuracy. A study involving the prediction of surface roughness in red brass turning was developed using a comparison between a complete factorial planning and ANN. The finding was that, based on ANN percentage error and the regression model, the artificial neural network was considered more accurate than the regression model [38]. Authors in [39] developed a model to investigate C23000 turning cutting parameter effects. Artificial neural networks and multiple regression models were used to model cutting forces based on cutting parameters, using variance analysis. It was concluded that the ANN model was more accurate than the regression model. Another study compared the mathematical regression with ANN, in order to predict the wear of the AISI 304 cutting tool. Experimental design includes four factors at five levels. According to the authors, the artificial neural network model was capable of better tool wear predictions [40]. A tool wear prediction by artificial neural networks in the Aluminum 7075 milling was proposed by [41]. Authors subdivided the research using the Taguchi model and ANNs. It was found that ANN prediction correlates very well with experimental results. In a turning paper, DOE was employed to select factor levels that benefit network prediction skills, concluding that the DOE methodology constitutes a better approach to the design of RBF networks for roughness prediction than most trial and error approaches. However, a frequently cited ANN disadvantage is the lack of a systematic way to design high-performance networks [35]. Another study deals with AISI 1040 steel machining. Response surface and ANN approaches were used. According to the authors, both approaches predict surface roughness accurately [42]. The authors in [43] reviewed several publications dealing with modeling and roughness by ANNs in machining processes. The review showed that most of the paper had a roughness (Ra) average prediction, and that little attention was given to the efficiency of the training. Researchers point out the definition of the ideal network topology as the main problem in roughness modeling. Optimization efforts are detected in a small number of publications, and comparisons between topology definition approaches are hardly found. Moreover, with regard to validation, most publications neglect or make this ability unclear. The use of the third validation data set can be found in only a few studies. The use of statistical evaluation to compare trained networks can be found in only one fifth of the peer-reviewed papers, in addition to the lack of a statistical evaluation comparing ANN-based models and models obtained by other methods. According to authors, the ANNs models´ accuracy are points that require more attention and many papers are presented only in graphic forms, thus lacking information as to the results’ reproduction. There is no standard procedure for choosing more appropriate ANN settings, thus becoming a difficult task that depends on many variables. The trial and error method is the procedure most commonly used to identify the best settings [44]. Based on the peer-reviewed literature, some areas of improvement are suggested, such as the no need to transform or change data, that reveal non-regular periodic movement. In addition, papers focus on network characteristics in a specification phase, with tests being performed to determine errors and R oscillations, and to validate a performance optimization model [45].
The objective of this paper is to evaluate the application of different MQL strategies in Ti-6Al-4V alloy machining, and to study whether the prediction of results between the ANNs approaches and factorial design describe similar values. The classical DOE approach and its comparison with a high number of different neural network architectures have not yet been studied. The number of comparative analysis studies between the DOE and ANN approaches is still limited. This paper includes training and testing of 15,000 artificial neural networks to minimize a systematic trial and error in the top milling of a Ti-6Al-4V alloy. An MQL valve was built for this purpose.

2. Materials and Methods

This was divided into two analyses: the first uses a factorial planning and its statistical correlations, and the second analysis uses artificial neural networks. Milling tests were performed using three different strategies. As results of the experimental planning, the influence of the independent variables on force and roughness parameters was verified by Pareto graphs. Statistical predictions were made using the generated response surfaces. The measured cutting and roughness data served to feed the artificial neural networks, and the predictions obtained by the two approaches were compared by means of a variance analysis. The desirability optimization function was used to seek optimal results aiming at lower roughness and lower cutting efforts. Figure 1 shows the experiment overview.
The strategies and materials utilized to manufacture the MQL valve, milling trials, and training of the ANNs are described in this section. A commercial Ti-6Al-4V (Grade 5) titanium alloy bar of 500 mm × 50 mm × 17 mm with an average hardness of 32.4 Hardness Rockwell C was used in this study. Figure 2 illustrates the material with the machining tests performed. Table 1 shows the respective chemical composition.

2.1. Experimental Design and Machining

An experimental design of three factors and two levels (23) with two central points was used, as shown in Table 2. A replication was performed for each cutting condition, totaling 60 tests. From these, the respective averages were calculated. The data adopted for the cutting speed (vc), feed rate (f), and depth of cut (ap) were recommended by the insert manufacturer. The procedure for using each cutting insert was used up to a wear of 0.1 mm.
The speed, feed rate, and depth of cut were considered as independent variables. As dependent variables, cutting force (Fc), feed force (Ff), penetration force (Fp), torque (Mz), root mean square height of the surface (Sq), Skewness of height distribution (Ssk) and Kurtosis of height distribution (Sku) were considered. Response surfaces were generated, and the desirability function was used to optimize the results of each strategy employed. Three different conditions were stipulated for the trials: (a) using no lubrication, called Strategy 1, (b) applying the designed mixing valve and adopting the MQL approach with oil (Strategy 2), and (c) using the MQL approach with the addition of 30% commercial graphite (Strategy 3). The trials were carried out according to Table 3. For the MQL approach, the Superfluid 3 lubricating oil from Quimatic/Tapmatic was applied. Its characteristics are listed in Table 4. The fluid was driven by a Magneti Marelli MAM00103 electric automotive pump (Mauá, Brazil), with a working pressure of 3 bar and a flow rate of 110 L/h connected to a 20 L fuel tank, and with an oil return circuit to the tank. The flow regulation was performed by a manual control valve set to 110 mL/h. The compressed air pressure was maintained at 6 bar.
The ROMI D600 machining center (Santa Bárbara d’Oeste, Brazil) with the SIEMENS 840-D CNC control (Figure 3a), a 20 hp motor, and total installed power of 30 kVA was used for milling tests. The insert BDMT 11 03 02 ER-JS (CVD) with grade CA6535 from KYOCERA (Sorocaba, Brazil) was chosen for milling (Figure 3b).
The MQL valve circuit (Sao Caetano do Sul, Brazil) is illustrated in Figure 4. The needle is connected to the bottom part of hydraulic equal cross coupling through a hose, and directs the flow of the mixture.
The needle and jet of fluid can be seen in Figure 5. The CAD design and the assembled valve can be seen in Figure 6. The valve was attached to a lubrication hose that was adequate to the machining center.

2.2. Measurements

For the measurement of forces during machining, a KISTLER 9123C rotary dynamometer (Sao Caetano do Sul, Brazil) with 5223B signal conditioner (Sao Paulo, Brazil) was used. Figure 7a shows the dynamometer mounted on the machine, and Figure 7b shows the tool and the mounted valve. The acquisition of the signals was carried out by the PCI 4472 card (Sao Caetano do Sul, Brazil), together with the Labview Signal Express 2.5 software from National Instruments. The frequency for data acquisition was 100 kHz. A Leica S6D microscope (Sao Paulo, Brazil) was used to check the wear after each machining, guaranteeing a Vb flank wear of up to 0.2 mm. Figure 8 shows the measurement of an insert.
The Talysurf CCI Lite white light interferometer (Sao Paulo, Brazil) was used for surface measurement. The surface roughness parameters generated using the software TalyMap Gold v. 6.2 and considered in this work were: Sq (root mean square height of the surface), Ssk (skewness of height distribution) and Sku (kurtosis of height distribution). The measuring area was 4.3mm2 with used 0.8mm cut-off. Figure 9 illustrates the measurement procedure.

2.3. Artificial Neural Networks

From the machining data obtained by the experimental design and measurements, artificial neural networks were created for each cutting condition. With the “Automated Neural Network” workbench from Statistica software (version 13.5), multilayer perceptron networks (MLP) in three independent variables as input were created, and with the machining forces and roughness measured as the “target.” Table 5 indicates the independent and dependent variables’ selection of the networks used.
The MLP architecture has a feedforward characteristic, with a one-way sequence between layers. Neurons have activation functions that use input signals of the anterior layer in a mathematical function, and the choice of activation function influences the behavior of the neural network. The most commonly used ANN type is feedforward, with input, hidden, and output layers. Network data are entered through the input layer and each neuron in this layer relates to each input variable. Hidden layers transmit input data to output layers by accurately simulating the function of the original system. The output layer is related to the results of the network, based on the predicted results errors and the weights of the connections values between neurons. The function of each neuron contains synapses, sums function, and transfer function. Through connections, neurons communicate, multiplying the input value with weight coefficients. These signals are transformed by an activation function, generating the network output result [46]. For each machining strategy, 5000 networks were trained, using various combinations of neuron numbers, activation functions and number of layers, until they reached satisfactory values. Table 6 shows the values used for the training, testing, and validation data for each chosen strategy.
Various training and tests were carried out using the test averages to validate the networks’ efficiency. Among the averages, three samples were randomly chosen for the tests, and seven samples were used for training. They were replicated 15 times, totaling 112 samples for training. The networks chosen were the ones that obtained the best results in training and testing, as exemplified in Figure 10.
The minimum and maximum number of neurons in the hidden layer were 3 and 25, respectively. The error function chosen was the sum of squares (SOS), due to the fact that it is one of the most used in training of artificial neural networks in regression tasks. The weight decay factor was stipulated between 0.0001 and 0.001. All available activation functions (identity, logistic, tanh, exponential and sine) were used for training.

2.4. Comparative Analysis

With the tabulated data, statistical tests were performed to verify the homogeneity of variances and to compare the pairs of averages. At the end of any experimental design, it should be checked whether the mathematical model obtained is valid [47]. Statistical tests were used to determine whether the two prediction strategies combined with the measurement produce equivalent estimates, i.e., to determine if there is a difference between the averages of results of the artificial neural networks and the experimental design. Analysis of variance was used, and therefore it can be seen if there are statistical differences between the predictions.
In addition, the desirability optimization function was used, in order to obtain optimal values for the three cutting conditions. The approach used in the desirability optimization consists of converting each response of the experimental design yi into a function di, varying between 0 and 1. If the answer is the desired one, a value of di = 1 is accepted, otherwise, di = 0 [47]. In this way, the independent variables are chosen to maximize the global desirability function. In this work, di = 1 was chosen as the lowest value for machining forces and roughness.

3. Results and Discussion

This topic provides the experimental design used, and the results of the average measurements of the machining forces and roughness and of the influence of the machining parameters, in relation to the dependent variables. The architectures of the ANNs are demonstrated, and a comparison of the treatments used is made.

3.1. Statistical Results of Surface Roughness Measurements

For every cutting condition, the maximum and minimum conditions of the parameters used were analyzed. The Sq parameter corresponds to the standard deviation of the distribution curve [48]. The Sku parameter is always presented with the Ssk parameter, describing the shape of a topographic surface and its roughness distribution. Mathematically, the parameters skewness and kurtosis measure the symmetry and histogram deviation of all peaks and valleys heights of a machined surface in relation to the Gaussian distribution. The skewness roughness parameter can monitor machined surfaces’ wear and tear in service. A surface with a Gaussian distribution, which is symmetrically distributed, has the Ssk parameter equal to zero. Positive Ssk values indicate the predominance of high peaks, and negative values indicate the prevalence of valleys. On the other hand, Sku parameter measures the degree of flattening or thinning of a topographic distribution of a peak´s roughness profile. On a surface with normal symmetric distribution, the Sku parameter is equal to three. In practical terms, Sku > 3 indicates the presence of acute peaks, while Sku < 3 indicates surface texture free of disproportionately sharp peaks [49,50]. In this work, the lowest Sq value was obtained by the number 2 assay used in strategy 3 (0.39 µm). The maximum condition occurred in test 10 strategy 1 (1.57 µm). For the Ssk parameter, the minimum value occurred in test 1 strategy 3 (−1.05), and the maximum condition occurred in test 1 strategy 2 (0.49). As for the Sku parameter, the minimum and maximum results are respectively found in test 2 strategy 1 (2.68) and test 1 strategy 3 (15.30). The statistical results are shown in Table 7. Based on statistical analysis results, the minimum and maximum topography surface values measured for each Sq, Ssk, Sku parameter are shown in Figure 11.

3.2. Results and Analysis of Factorial Design

The use of experimental design resulted in measurements of the dependent variables for the three lubrication conditions. The data can be viewed in Table 8. The analysis of the feed rate, cutting, and penetration machining forces for the three tests, as illustrated in the following items, showed that these forces are dominated by the feed rate and the penetration depth variables. It is known that the greater these parameters are, the greater the chip removal forces are and, consequently, the greater the forces generated to remove them are. It has been noted that the cutting force decreases strongly in applications with oil and oil + graphite in relation to the cut without using fluid, due to the fluid emulsivity, minimizing friction, and temperature, acting directly on the flank of the cutting insert. The type of base fluid has a strong effect on the components of the cutting force. Consequently, the force is significantly reduced [1]. For the roughness parameter, the main factors that influence it are the feed rate, the cutting speed, the depth of cut, the vibrations of the cutting tools, the lubricant conditions, and the tool specifications [19]. In this study, it was found that the feed rate and ap were the variables that most influenced roughness.

3.2.1. Trial without Lubrication

Based on the Pareto graphs, it was observed that, for a 95% confidence interval, the cutting speed, advancing and forces were not influenced by the dependent variables. The roughness parameters also did not show any influence on the confidence interval adopted, although the ap and f are the variables with the greatest influence. Analyzing the illustration in Figure 12, the torque was influenced by the feed rate f and the depth of cut ap.
In analyzing the response surfaces obtained by the factorial design of the test without lubrication illustrated in Figure 13, it can be seen that the minimum torque values occur with lower feeds, and lower depths of cut. The analysis process discards the statistical value of vc, since it does not influence the torque in the studied confidence interval. The lowest values of cutting forces, feed rate force, penetration force, and torque occur in smaller feeds, smaller depths of cut, and lower cutting speeds. The contributions of ap and f are striking, indicating that they are possibly responsible for the mechanism regulating the machining forces of this test condition. The area of the section of cut that is defined by the product of the feed by the depth of cut increases with the increase of the feed and the depth of cut, causing an increase in the forces in question.

3.2.2. Trial with Oil

Based on the Pareto graphs illustrated in Figure 14, it was observed that the variable ap, the advance f, and the interaction ap and f demonstrated an influence on the cutting force. The variables ap and f demonstrated an influence on the advance force for the 95% confidence interval.
For the same confidence interval, the penetration force was not influenced by the independent variables. Advancement was the most influential variable when a lower confidence interval was adopted.
The torque was influenced by the depth of cut and the feed. For the studied interval, the roughness Sq was influenced by interaction ap and f. The Ssk parameter was influenced by the cutting speed, the feed rate and the interactions between vc and f and vc and ap.
Analyzing the response surfaces obtained by the test´s factorial with lubrication, it can be seen that the minimum Sq values also occur for lower f and higher ap.
The lowest values of cutting forces and torque, illustrated in Figure 15, occur with smaller feed rates and smaller depths of penetration—discarding the cutting speed since it does not influence the approached parameter. The smallest feed rate forces, penetration forces, and torque occur with smaller depths of cut and smaller feed rates.

3.2.3. Trial with Graphite

Based on the Pareto graphs illustrated in Figure 16, it was observed that the variable ap and the feed rate showed an influence on the cutting force. The ap variable showed an influence on the feed rate force, and for a 95% confidence interval, the penetration force was not influenced by the independent variables. The depth of cut was the variable of greatest influence for the penetration force. The torque was influenced by the depth of penetration and the feed rate. For the confidence interval adopted, the Sq roughness was influenced by the feed rate.
Analyzing the response surfaces obtained by the factorial design of the trial with oil and graphite illustrated in Figure 17, it can be seen that the minimum roughness values occur with lower feed rates and smaller depths of cut. The lower values of the cutting forces, feed rate forces, penetration forces, and torque occur with smaller feed rates, smaller depths of penetration, and lower cutting speeds.
Table 9 lists the functions of the responses of the variables Fc, Ff, Fp, Mz, Ssk, Sku and Sq, obtained by the application of the factorial plan in this work.
In tests with no lubrication, it can be perceived that the machining forces (Fc, Ff, Fp) are not statistically influenced by the independent variables. Only torque was influenced by the parameters f and ap in the confidence interval studied. In the tests with lubrication, both with oil and graphite, it is apparent that machining forces and torque are significantly influenced by the parameters ap and f. The Sq roughness was influenced by the interaction between ap and f in the oil assay, while in the graphite assay, only the f parameter showed significant influence.
The roughness increased with higher feed rates. With the cut depth and feed increase, the cutting area in the tool is increased and, with this, the cutting effort is increased. In the oil strategy, the minimum Ssk values occur with higher cutting speeds and greater feed rates. This is due to a reduction in plastic deformation in the cutting zone incurred by a higher cutting speed. The surface defect is smaller, with a lower overall roughness value. Surface roughness decreases with increased cutting speed due to lower plastic deformation in the cutting zone associated with high speeds [51]. With the increase of the cutting speed, the temperature is increased, and the force is minimized by heating the material, helping its removal, and reducing roughness that, in turn, depends on feed rates. Oil application contributes to this process, minimizes temperature and favors tool life. Tool wear and coolant use generally influence surface finish [52]. This can be seen with the lowest roughness values in the strategies using oil and graphite-added oil. Lower cutting forces and lower feed rates provided lower roughness values. The analysis infers that the cutting force corroborates the surface roughness analysis [52].

3.2.4. Machined Surface Roughness Analysis

Roughness parameters were shown to be governed by the independent variables ap and f. The variable vc is only noted in this work influencing Ssk in Strategy 2. To verify the roughness parameters behavior studied, the study of the lowest and highest values of the cut-off conditions was adopted as a premise. Thus, tests 1 and 8 of each experimental planning strategy are analyzed. Figure 18 shows the isometric surfaces and contour maps of the tests for the dry strategy. An increase of Sq in 99.8% was perceived due to the increase in ap and f causing greater irregularities of the roughness profile. The negative Ssk parameter for both conditions indicates the prevalence of valleys. The positive Sku parameter indicates that the peaks of the machined surface are sharp.
Figure 19 shows the isometric surfaces and contour maps of superfluid tests. The Ssk parameter, positive for the minimum cut-off condition, indicates the prevalence of peaks, which does not occur for the maximum values. The positive Sku parameter indicates that the peaks of the machined surface are sharp. It is observed that the increase in feed and cutting speed accentuated the milling marks.
Figure 20 shows the isometric surfaces and contour maps of the tests with the mixture of Superfluid 3 with graphite. An increase of Sq of 27% is noted. The Ssk parameter, negative for both cutting conditions, indicates the prevalence of valleys. The positive Sku parameter indicates that the peaks of the machined surface are sharp. It is observed that the increase of f accentuated the advance marks.

3.2.5. Desirability Function

Once you have the answer functions for all variables dependent on the research, an optimization bringing together all the variables allows you to show the best values for the application of the independent variables. The desirability function is used in single and multiple objective functions. Unlike other strategies, this optimization function avoids a conflict of responses. The variation of the input parameters must be planned to allow for working within the specified experimental design interval [17]. Using the Statistica software, optimal values were sought using the desirability function for each dependent variable. Table 10 and Figure 21 illustrate the results.

3.3. Prediction of Results by Artificial Neural Networks

With the data obtained by measuring the machining forces and roughness, different training standards and trials of the artificial neural networks were performed. Trainings were performed for different lubrication conditions with an average of 10 samples, creating supervised networks with three input variables (vc, f, and ap). As the target, the dependent samples Ff, Fc, Fp, Mz, and Ra were selected.
For the execution of the training and tests, the Automated Neural Network tool of the Statistica software was used. The tests represent the result of the simulation of the previously trained networks. For a future test to validate the training, three samples were randomly reserved. The remaining 7 samples were replicated 15 times, totaling 112 samples for training. Table 11 shows the variables chosen randomly for the validation tests. Table 12 contains information on the architectures of the artificial neural networks used to predict.
The test results can be seen in Table 13 and Table 14. The measured values and the results of the ANNs are listed respectively as output. There is a good approximation between the values.

3.4. Comparative Analysis of Predictions

To verify whether the prediction strategies used in this work produce similar responses, a statistical test was carried out to verify the equality of treatment averages (ANOVA). It was found, as shown in Figure 22, that there were no global differences between prediction strategies (p-value > 0.05). In these graphs, the numbers 1, 2, and 3 represent the measured values, the results of the experimental design using the equations, and the results of the ANNs, respectively.
Future research can use statistical tests such as Duncan, Bonferroni, Tukey, and the validation in the case of differences in averages that they indicate, can be verified by the Levene test. The inclusion of other percentages of graphite and its comparison with other lubrication strategies, such as liquid nitrogen under dripping, would be interesting.

4. Conclusions

The design of the MQL mixing valve made it possible to study and analyze different lubrication conditions. The assembled circuit was based on commercial parts. It was found that the independent variables that most influenced the cutting forces and roughness were the depth of cut and feed rate. The cutting forces were significantly reduced with the use of the MQL approach, probably allowing for heat reduction in the cutting flank of the insert, due to the fluid emulsivity. The lowest average of the Sq roughness was achieved by using the MQL practice with cutting fluid and graphite. The desirability function provided the optimized search for the lowest cutting forces and for a lower roughness value, aiming at the surface finish. In tests with no lubrication, it can be perceived that the machining forces (Fc, Ff, Fp) are not statistically influenced by the independent variables. Only torque was influenced by the parameters f and ap in the confidence interval studied. In the tests with lubrication, both with oil and graphite, it is apparent that machining forces and torque are significantly influenced by the parameters ap and f. The Sq roughness was influenced by the interaction between ap and f in the oil assay, while in the graphite assay, only the f parameter showed significant influence. Lower cutting forces and lower feed rates provided lower roughness values. The paper took into account surface finishing cutting conditions. In the lowest feed rate and depth cut, it can be perceived that statistically, for the confidence interval adopted, that the forces had lower statistical influences than variables ap and f. It has been observed that variables ap and f define the cutting geometry, ergo being responsible for the roughness. The Ssk values showed values close to zero in most tests, indicating a symmetry between peaks and valleys. Negative values, however, indicate a trend of valley predominance and a concentration of material near the surface. All Sku parameters measured demonstrated positive values, showing that the tests presented centralized acute peaks. With sharp peaks, there is the possibility of premature wear and tear in contact with another surface. The statistical comparison between the experimental design adopted and the artificial neural networks provided similar responses.

Author Contributions

Conceptualization, N.W.P., E.C.B., and G.F.B.; methodology, N.W.P., A.F.d.L.F. and J.A.G.F.; software, N.W.P., G.d.L.X.R., and C.C.; validation, E.C.B., and J.A.G.F.; formal analysis, G.F.B. and J.A.G.F.; investigation, N.W.P., A.F.d.L.F. and J.A.G.F.; resources, N.W.P., A.F.d.L.F. and E.C.B.; writing—original draft preparation, N.W.P. and E.C.B.; writing—review and editing, N.W.P., E.C.B. and J.A.G.F.; visualization, G.d.L.X.R.; supervision, G.F.B. All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Acknowledgments

SENAI-SP and Instituto Mauá de Tecnologia are acknowledged for making their laboratories available for the execution of the tests.

Conflicts of Interest

The authors declare no conflict of interest.

Nomenclature

ANNArtificial Neural Networks
ANOVAAnalysis of variance
ap Depth of cut (mm)
BPNNBackpropagation neural network
CVDChemical vapor deposition
CNCComputer numeric controlled
DOEDesign of experiments
f Feed rate (mm/rev)
FcCutting force (N)
FfFeed force (N)
FpPenetration force (N)
GAGenetic algorithm
MQLMinimum quantity lubrication
MzTorque (Nm)
MLPMulti Layer Perceptron
SkuKurtosis of height distribution (dimensionless)
SOSSum of squares error function (dimensionless)
SskSkewness of height distribution (dimensionless)
SqRoot mean square height of the surface (µm)
vc Cutting speed (m/min)
VbFlank wear (mm)
2(**)two levels

References

  1. Duc, T.M.; Tran, T.L.; Tran, Q.C. Performance evaluation of MQL parameters using Al2O3 and MoS2 nanofluids in hard turning 90CrSi steel. Lubricants 2019, 7, 40. [Google Scholar] [CrossRef] [Green Version]
  2. Rodríguez, J.M.; Larsson, S.; Carbonell, J.M.; Jonsén, P. Dislocation density based flow stress model applied to the PFEM simulation of orthogonal cutting processes of Ti-6Al-4V. Materials 2020, 13, 1979. [Google Scholar] [CrossRef]
  3. Roy, S.; Kumar, R.; Sahoo, A.K.; Das, R.K. A brief review on effects of conventional and nano particle based machining fluid on machining performance of minimum quantity lubrication machining. Mater. Today Proc. 2019, 18, 5421–5431. [Google Scholar] [CrossRef]
  4. Nagaraj, P.B.; Dinesh, P. Friction force during machining process—Part 1: Development of optimized neural network architecture. Mater. Today Proc. 2020, 27, 1407–1410. [Google Scholar] [CrossRef]
  5. D’Addona, D.M.; Raykar, S.J. Thermal modeling of tool temperature distribution during high pressure coolant assisted turning of Inconel 718. Materials 2019, 12, 408. [Google Scholar] [CrossRef] [Green Version]
  6. García-Martínez, E.; Miguel, V.; Martínez-Martínez, A.; Manjabacas, M.C.; Coello, J. Sustainable Lubrication Methods for the Machining of Titanium Alloys: An Overview. Materials 2019, 12, 3852. [Google Scholar] [CrossRef] [Green Version]
  7. Shokrani, A.; Newman, S.T. A new cutting tool design for cryogenic machining of Ti–6Al–4V titanium alloy. Materials 2019, 12, 477. [Google Scholar] [CrossRef] [Green Version]
  8. Shokrani, A.; Al-Samarrai, I.; Newman, S.T. Hybrid cryogenic MQL for improving tool life in machining of Ti-6Al-4V titanium alloy. J. Manuf. Process. 2019, 43, 229–243. [Google Scholar] [CrossRef]
  9. Abbas, A.T.; Sharma, N.; Anwar, S.; Luqman, M.; Tomaz, I.; Hegab, H. Multi-response optimization in high-speed machining of Ti-6Al-4V using TOPSIS-fuzzy integrated approach. Materials 2020, 13, 1104. [Google Scholar] [CrossRef] [Green Version]
  10. Abbas, A.T.; Sharma, N.; Anwar, S.; Hashmi, F.H.; Jamil, M.; Hegab, H. Towards optimization of surface roughness and productivity aspects during high-speed machining of Ti–6Al–4V. Materials 2019, 12, 3749. [Google Scholar] [CrossRef] [Green Version]
  11. Sánchez Hernández, Y.; Trujillo Vilches, F.J.; Bermudo Gamboa, C.; Sevilla Hurtado, L. Experimental parametric relationships for chip geometry in dry machining of the Ti6Al4V alloy. Materials 2018, 11, 1260. [Google Scholar] [CrossRef] [Green Version]
  12. Qin, S.; Li, Z.; Guo, G.; An, Q.; Chen, M.; Ming, W. Analysis of minimum quantity lubrication (MQL) for different coating tools during turning of TC11 titanium alloy. Materials 2016, 9, 804. [Google Scholar] [CrossRef] [PubMed]
  13. Liu, D.; Zhang, Y.; Luo, M.; Zhang, D. Investigation of tool wear and chip morphology in dry trochoidal milling of titanium alloy Ti–6Al–4V. Materials 2019, 12, 1937. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  14. Matras, A.; Zębala, W.; Machno, M. Research and method of roughness prediction of a curvilinear surface after titanium alloy turning. Materials 2019, 12, 502. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  15. Fernández-Pérez, J.; Cantero, J.L.; Díaz-Álvarez, J.; Miguélez, M.H. Hybrid composite-metal stack drilling with different minimum quantity lubrication levels. Materials 2019, 12, 448. [Google Scholar] [CrossRef] [Green Version]
  16. Słodki, B.; Zębala, W.; Struzikiewicz, G. Turning titanium alloy, grade 5 ELI, with the implementation of high pressure coolant. Materials 2019, 12, 768. [Google Scholar] [CrossRef] [Green Version]
  17. Singh, G.; Pruncu, C.I.; Gupta, M.K.; Mia, M.; Khan, A.M.; Jamil, M.; Pimenov, D.Y.; Sen, B.; Sharma, V.S. Investigations of machining characteristics in the upgraded MQL-assisted turning of pure titanium alloys using evolutionary algorithms. Materials 2019, 12, 999. [Google Scholar] [CrossRef] [Green Version]
  18. Gupta, M.K.; Jamil, M.; Wang, X.; Song, Q.; Liu, Z.; Mia, M.; Hegab, H.; Khan, A.M.; Collado, A.G.; Pruncu, C.I.; et al. Performance evaluation of vegetable oil-based nano-cutting fluids in environmentally friendly machining of Inconel-800 alloy. Materials 2019, 12, 2792. [Google Scholar] [CrossRef] [Green Version]
  19. James, S.J.; Annamalai, A.R. Machinability study of developed composite AA6061-ZrO2 and analysis of influence of MQL. Metals 2018, 8, 472. [Google Scholar] [CrossRef] [Green Version]
  20. Mia, M.; Morshed, M.; Kharshiduzzaman, M.; Razi, M.H.; Mostafa, M.R.; Rahman, S.M.; Ahmad, I.; Hafiz, M.T.; Kamal, A.M. Prediction and optimization of surface roughness in minimum quantity coolant lubrication applied turning of high hardness steel. Measurement 2018, 118, 43–51. [Google Scholar] [CrossRef]
  21. Abbas, A.T.; Benyahia, F.; El Rayes, M.M.; Pruncu, C.; Taha, M.A.; Hegab, H. Towards optimization of machining performance and sustainability aspects when turning AISI 1045 steel under different cooling and lubrication strategies. Materials 2019, 12, 3023. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  22. Dong, P.Q.; Duc, T.M.; Long, T.T. Performance evaluation of MQCL hard milling of SKD 11 tool steel using MoS2 nanofluid. Metals 2019, 9, 658. [Google Scholar] [CrossRef]
  23. Sousa, V.F.C.; Silva, F.J.G. Recent advances in turning processes using coated tools—A comprehensive review. Metals 2020, 10, 170. [Google Scholar] [CrossRef] [Green Version]
  24. Sankar, M.R.; Saxena, S.; Banik, S.R.; Iqbal, I.M.; Nath, R.; Bora, L.J.; Gajrani, K.K. Experimental study and artificial neural network modeling of machining with minimum quantity cutting fluid. Mater. Today Proc. 2019, 18, 4921–4931. [Google Scholar] [CrossRef]
  25. Swain, S.; Panigrahi, I.; Kumar Sahoo, A.; Panda, A. Study on machining performances during hard turning process using vibration signal under MQL environment: A review. Mater. Today Proc. 2019, 18, 3539–3545. [Google Scholar] [CrossRef]
  26. Vishnu Vardhan, M.; Sankaraiah, G.; Yohan, M. Optimization of cutting parameters and prediction of Ra & MRR for machining of P20 Steel on CNC milling using Artificial Neural Networks. Mater. Today Proc. 2018, 5, 27058–27064. [Google Scholar] [CrossRef]
  27. Vardhan, M.V.; Sankaraiah, G.; Yohan, M. Prediction of surface roughness & material removal rate for machining of P20 steel in CNC milling using artificial neural networks. Mater. Today Proc. 2018, 5, 18376–18382. [Google Scholar] [CrossRef]
  28. Paturi, U.M.; Devarasetti, H.; Narala, S.K. Application of regression and artificial neural network analysis in modelling of surface roughness in hard turning of AISI 52100 steel. Mater. Today Proc. 2018, 5, 4766–4777. [Google Scholar] [CrossRef]
  29. Prasad, M.D.; Malyadri, T.; Hari, S.S. Sensitivity analysis for process parameters influencing surface roughness of hardened steel in dry machining process. Mater. Today Proc. 2020, 26, 2521–2524. [Google Scholar] [CrossRef]
  30. Arnold, F.; Hänel, A.; Nestler, A.; Brosius, A. New approaches for the determination of specific values for process models in machining using artificial neural networks. Procedia Manuf. 2017, 11, 1463–1470. [Google Scholar] [CrossRef]
  31. Kant, G.; Sangwan, K.S. Predictive modelling and optimization of machining parameters to minimize surface roughness using artificial neural network coupled with genetic algorithm. Procedia CIRP 2015, 31, 453–458. [Google Scholar] [CrossRef] [Green Version]
  32. Oneto, L.; Bunte, K.; Schleif, F.-M. Advances in artificial neural networks, machine learning and computational intelligence: Selected papers from the 26th European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning (ESANN 2018). Neurocomputing 2019, 342, 1–5. [Google Scholar] [CrossRef]
  33. Pontes, F.J.; Amorim, G.F.; Balestrassi, P.P.; Paiva, A.P.; Ferreira, J.R. Design of experiments and focused grid search for neural network parameter optimization. Neurocomputing 2016, 186, 22–34. [Google Scholar] [CrossRef]
  34. Lasheras, F.S.; Vilán, J.V.; Nieto, P.G.; del Coz Díaz, J.J. The use of design of experiments to improve a neural network model in order to predict the thickness of the chromium layer in a hard chromium plating process. Math. Comput. Model. 2010, 52, 1169–1176. [Google Scholar] [CrossRef]
  35. Pontes, F.J.; Silva, M.B.; Ferreira, J.R.; Paiva, A.P.D.; Balestrassi, P.P.; Schönhorst, G.B. A DOE based approach for the design of RBF artificial neural networks applied to prediction of surface roughness in AISI 52100 hardened steel turning. J. Braz. Soc. Mech. Sci. Eng. 2010, 32, 503–510. [Google Scholar] [CrossRef]
  36. Kumar, D.; Chandna, P.; Pal, M. Efficient Optimization of Neural Network using Taguchi-grey Relational Analysis with Signal-to-Noise Ratio Approach for 2.5 D End Milling Process. Am. J. Mech. Eng. Autom. 2018, 5, 30. [Google Scholar]
  37. Morales Tamayo, Y.; Beltrán Reyna, R.F.; López Bustamante, R.J.; Zamora Hernández, Y.; López Cedeño, K.; Terán Herrera, H.C. Comparison of two methods for predicting surface roughness in turning stainless steel AISI 316L. Ingeniare Rev. Chil. Ing. 2018, 26, 97–105. [Google Scholar] [CrossRef] [Green Version]
  38. Hanief, M.; Wani, M.F. Artificial neural network and regression-based models for prediction of surface roughness during turning of red brass (C23000). J. Mech. Eng. Sci. 2016, 10, 1835–1845. [Google Scholar] [CrossRef]
  39. Hanief, M.; Wani, M.F.; Charoo, M.S. Modeling and prediction of cutting forces during the turning of red brass (C23000) using ANN and regression analysis. Eng. Sci. Technol. Int. J. 2017, 20, 1220–1226. [Google Scholar] [CrossRef] [Green Version]
  40. Kalidass, S.; Palanisamy, P.; Muthukumaran, V. Prediction of tool wear using regression and artificial neural network models in end milling of AISI 304 austenitic stainless steel. Int. J. Eng. Innov. Technol. 2012, 1, 29–35. [Google Scholar]
  41. Khorasani, A.M.; Yazdi, M.R.S.; Safizadeh, M.S. Tool Life Prediction in Face Milling Machining of 7075 Al by Using Artificial Neural Networks (ANN) and Taguchi Design of Experiment (DOE). Int. J. Eng. Technol. 2011, 3, 30. [Google Scholar] [CrossRef] [Green Version]
  42. Sahoo, A.; Rout, A.; Das, D. Response surface and artificial neural network prediction model and optimization for surface roughness in machining. Int. J. Ind. Eng. Comput. 2015, 6, 229–240. [Google Scholar] [CrossRef]
  43. Pontes, F.J.; Ferreira, J.R.; Silva, M.B.; Paiva, A.P.; Balestrassi, P.P. Artificial neural networks for machining processes surface roughness modeling. Int. J. Adv. Manuf. Technol. 2010, 49, 879–902. [Google Scholar] [CrossRef]
  44. Kechagias, J.; Tsiolikas, A.; Asteris, P.; Vaxevanidis, N. Optimizing ANN performance using DOE: Application on turning of a titanium alloy. MATEC Web Conf. 2018, 178, 1017. [Google Scholar] [CrossRef]
  45. Abiodun, O.I.; Jantan, A.; Omolara, A.E.; Dada, K.V.; Mohamed, N.A.; Arshad, H. State-of-the-art in artificial neural network applications: A survey. Heliyon 2018, 4, e938. [Google Scholar] [CrossRef] [Green Version]
  46. Karkalos, N.E.; Galanis, N.I.; Markopoulos, A.P. Surface roughness prediction for the milling of Ti–6Al–4V ELI alloy with the use of statistical and soft computing techniques. Measurement 2016, 90, 25–35. [Google Scholar] [CrossRef]
  47. Calado, V.; Montgomery, D.C. Planejamento de Experimentos Usando o Statistica; Editora E-papers: Rio de Janeiro, Brazil, 2003; ISBN 978-85-8792-283-0. [Google Scholar]
  48. Unda, A.G.R.; Lin, V.F.C.; De Godoy, G.C.D. Metodologia para a aplicação da análise topográfica ao estudo de processos de superfície. Matéria 2007, 12, 589–596. [Google Scholar] [CrossRef] [Green Version]
  49. Dong, W.P.; Sullivan, P.J.; Stout, K.J. Comprehensive study of parameters for characterizing three-dimensional surface topography III: Parameters for characterizing amplitude and some functional properties. Wear 1994, 178, 29–43. [Google Scholar] [CrossRef]
  50. Rodrigues, A.R.; Manarelli, F.H.; De Queiroz, M.C.G.P.; Matsumoto, H.; Yamakami, W.J. Rugosidade e microestrutura da peça no fresamento do aço vp100 para moldes. In Proceedings of the 7th Brazilian Congress on Manufacturing Engineering, Penedo, Itatiaia, RJ, Brazil, 15–19 April 2013. [Google Scholar]
  51. Oosthuizen, G.A.; Nunco, K.; Conradie, P.J.T.; Dimitrov, D.M. The effect of cutting parameters on surface integrity in milling Ti6Al4V. S. Afr. J. Ind. Eng. 2016, 27, 115–123. [Google Scholar] [CrossRef] [Green Version]
  52. Polishetty, A.; Goldberg, M.; Littlefair, G.; Puttaraju, M.; Patil, P.; Kalra, A. A preliminary assessment of machinability of titanium alloy Ti6Al4V during thin wall machining using trochoidal milling. Procedia Eng. 2014, 97, 357–364. [Google Scholar] [CrossRef] [Green Version]
Figure 1. Experiment overview.
Figure 1. Experiment overview.
Materials 13 03828 g001
Figure 2. Ti-6AL-4V bar after machining.
Figure 2. Ti-6AL-4V bar after machining.
Materials 13 03828 g002
Figure 3. (a) ROMI D600 machining center and (b) KYOCERA tool used for milling.
Figure 3. (a) ROMI D600 machining center and (b) KYOCERA tool used for milling.
Materials 13 03828 g003
Figure 4. MQL valve circuit. (1) oil inlet; (2) return hose; (3) regulating valve; (4) check valve; (5) hydraulic equal cross coupling—air + oil mixture; (6) inlet compressed air; (7) control valve for compressed air.
Figure 4. MQL valve circuit. (1) oil inlet; (2) return hose; (3) regulating valve; (4) check valve; (5) hydraulic equal cross coupling—air + oil mixture; (6) inlet compressed air; (7) control valve for compressed air.
Materials 13 03828 g004
Figure 5. Needle and jet of Superfluido 3.
Figure 5. Needle and jet of Superfluido 3.
Materials 13 03828 g005
Figure 6. CAD design and the assembled valve.
Figure 6. CAD design and the assembled valve.
Materials 13 03828 g006
Figure 7. (a) The KISTLER rotary dynamometer. (b) the tool and the mounted valve.
Figure 7. (a) The KISTLER rotary dynamometer. (b) the tool and the mounted valve.
Materials 13 03828 g007
Figure 8. Wear measurement example.
Figure 8. Wear measurement example.
Materials 13 03828 g008
Figure 9. Roughness measurement procedure. (a) Taylor Hobson/Talysurf CCI Lite white light interferometer (b) Position and application area. (c) Measurement being performed. (d) Milled 3D surface topography.
Figure 9. Roughness measurement procedure. (a) Taylor Hobson/Talysurf CCI Lite white light interferometer (b) Position and application area. (c) Measurement being performed. (d) Milled 3D surface topography.
Materials 13 03828 g009
Figure 10. Scatterplot of Fc_Target x Fc_Output ANN.
Figure 10. Scatterplot of Fc_Target x Fc_Output ANN.
Materials 13 03828 g010
Figure 11. Sq, Ssk, Sku minimum and maximum milled surface topography parameter values measured. (a) Sq = 0.39 µm; (b) Sq = 1.57 µm; (c) Ssk = −1.05 and Sku = 15.30; (d) Ssk = 0.49 µm; (e) Ssk = 2.68 µm
Figure 11. Sq, Ssk, Sku minimum and maximum milled surface topography parameter values measured. (a) Sq = 0.39 µm; (b) Sq = 1.57 µm; (c) Ssk = −1.05 and Sku = 15.30; (d) Ssk = 0.49 µm; (e) Ssk = 2.68 µm
Materials 13 03828 g011
Figure 12. Pareto chart for the dry machining condition: torque Mz.
Figure 12. Pareto chart for the dry machining condition: torque Mz.
Materials 13 03828 g012
Figure 13. Response surfaces for the non-lubricated condition (Mz).
Figure 13. Response surfaces for the non-lubricated condition (Mz).
Materials 13 03828 g013
Figure 14. Pareto charts for oil machining condition. (a) Force Fc. (b) Force Ff. (c) Torque Mz. (d) Roughness Sq. (e) Skewness of height distribution.
Figure 14. Pareto charts for oil machining condition. (a) Force Fc. (b) Force Ff. (c) Torque Mz. (d) Roughness Sq. (e) Skewness of height distribution.
Materials 13 03828 g014
Figure 15. Response surfaces for the oil lubricated condition. (a) Cutting force Fc (ap × f). (b) Feed force Ff (ap × f). (c) Torque Mz (ap × f). (d) Skewness (f × vc).
Figure 15. Response surfaces for the oil lubricated condition. (a) Cutting force Fc (ap × f). (b) Feed force Ff (ap × f). (c) Torque Mz (ap × f). (d) Skewness (f × vc).
Materials 13 03828 g015
Figure 16. Pareto graphs for machining condition with graphite. (a) Cutting force Fc. (b) Feed force Ff. (c) Torque Mz. (d) Roughness Sq.
Figure 16. Pareto graphs for machining condition with graphite. (a) Cutting force Fc. (b) Feed force Ff. (c) Torque Mz. (d) Roughness Sq.
Materials 13 03828 g016
Figure 17. Response surfaces for the lubrication condition with graphite. (a) Cutting force Fc (ap × f). (b) Feed force Ff (ap × f). (c) Torque Mz (ap × f). (d) Roughness Sq (ap × f).
Figure 17. Response surfaces for the lubrication condition with graphite. (a) Cutting force Fc (ap × f). (b) Feed force Ff (ap × f). (c) Torque Mz (ap × f). (d) Roughness Sq (ap × f).
Materials 13 03828 g017
Figure 18. Machined surface topography. Isometric views and contour maps—dry strategy. (a)vc = 80 m/min, ap = 0.5 mm, f = 0.06 mm/rev.; (b) vc = 100 m/min, ap = 1.0 mm, f = 0.1 mm/rev.
Figure 18. Machined surface topography. Isometric views and contour maps—dry strategy. (a)vc = 80 m/min, ap = 0.5 mm, f = 0.06 mm/rev.; (b) vc = 100 m/min, ap = 1.0 mm, f = 0.1 mm/rev.
Materials 13 03828 g018
Figure 19. Machined surface topography. Isometric views and contour maps—oil strategy. (a)vc = 80 m/min, ap = 0.5 mm, f = 0.06 mm/rev.; (b) vc = 100 m/min, ap = 1.0 mm, f = 0.1 mm/rev
Figure 19. Machined surface topography. Isometric views and contour maps—oil strategy. (a)vc = 80 m/min, ap = 0.5 mm, f = 0.06 mm/rev.; (b) vc = 100 m/min, ap = 1.0 mm, f = 0.1 mm/rev
Materials 13 03828 g019
Figure 20. Machined surface topography. Isometric views and contour maps—oil + graphite strategy. (a) vc = 80 m/min, ap = 0.5 mm, f = 0.06 mm/rev.; (b) vc = 100 m/min, ap = 1.0 mm, f = 0.1 mm/rev.
Figure 20. Machined surface topography. Isometric views and contour maps—oil + graphite strategy. (a) vc = 80 m/min, ap = 0.5 mm, f = 0.06 mm/rev.; (b) vc = 100 m/min, ap = 1.0 mm, f = 0.1 mm/rev.
Materials 13 03828 g020
Figure 21. Desirability response surfaces.
Figure 21. Desirability response surfaces.
Materials 13 03828 g021
Figure 22. Differences between pairs of averages. (a) Cutting force Fc (b) Feed force Ff. (c) Penetration Force Fp. (d) Torque Mz. (e) Roughness Sq. (f) Skewness. (g) Kurtosis.
Figure 22. Differences between pairs of averages. (a) Cutting force Fc (b) Feed force Ff. (c) Penetration Force Fp. (d) Torque Mz. (e) Roughness Sq. (f) Skewness. (g) Kurtosis.
Materials 13 03828 g022aMaterials 13 03828 g022b
Table 1. Chemical composition of Ti-6Al-4V (provided by VINER Brasil Tecnologia Ltda., Sao Paulo, Brazil).
Table 1. Chemical composition of Ti-6Al-4V (provided by VINER Brasil Tecnologia Ltda., Sao Paulo, Brazil).
CompositionTiAlVFeHNOCY
Data (%)Balance6.49–6.564.03–4.140.16–0.190.002–0.0030.003–0.0040.192–0.1960.024–0.028<0.001
Table 2. Factorial design used for machining.
Table 2. Factorial design used for machining.
Testsvc (m/min)f (mm/rev)ap (mm)
1800.060.5
21000.060.5
3800.10.5
41000.10.5
5800.061.0
61000.061.0
7800.11.0
81000.11.0
9900.080.75
10900.080.75
Table 3. Lubrication strategies used (dry/oil/oil + graphite).
Table 3. Lubrication strategies used (dry/oil/oil + graphite).
StrategyCooling and Lubrication Condition
1Dry (only compressed air)
2Oil (minimum quantity lubrication (MQL))
3Oil + Graphite (MQL)
Table 4. Characteristics of the cutting fluid used (provided by QUIMATIC/TAPMATIC).
Table 4. Characteristics of the cutting fluid used (provided by QUIMATIC/TAPMATIC).
ParametersProperties
AspectClear green viscous liquid
ScentCharacteristic, light
Density at 25 °C (g/mL)0.88–0.91
Corrosion test on iron chipsNo corrosion after 2 h
Viscosity centistokes, 25 °C20–35
Acidity index10–20
Table 5. Select variables for analysis.
Table 5. Select variables for analysis.
Train, Test and Validation—ANNsVariables
Continuous inputsvc, f, ap
Continuous targetsFc, Ff, Fp, Mz, Sq, Ssk, Sku
Table 6. Sampling Method (Random).
Table 6. Sampling Method (Random).
Train sample size (%)70
Test sample size (%)20
Validation sample size (%)10
Table 7. Descriptive statistics—roughness.
Table 7. Descriptive statistics—roughness.
VariableMeanMinimumMaximumStd. Dev.
Sq (µm)0.690.391.570.37
Ssk−0.14−1.050.490.26
Sku4.352.6815.302.22
Table 8. Results of factorial design used.
Table 8. Results of factorial design used.
Dryvc (m/min)f (mm/rev)ap (mm)Fc (N)Ff (N)Fp (N)Mz (Nm)Sq (µm)SskSku
1800.060.5258.9771.0052.040.620.57−0.133.95
21000.060.5228.50108.0060.960.680.80−0.172.68
3800.10.5364.12112.9993.741.030.82−0.203.06
41000.10.5427.50109.00113.530.971.01−0.293.34
5800.061.0371.8094.7363.351.011.13−0.063.18
61000.061.0396.81100.5070.970.981.310.093.65
7800.11.0510.50172.62119.371.441.460.043.07
81000.11.0531.50139.40103.141.451.14−0.053.60
9900.080.75570.50164.42203.951.141.51−0.094.30
10900.080.75617.50207.94207.941.311.570.365.39
Oilvc (m/min)f (mm/rev)ap (mm)Fc (N)Ff (N)Fp (N)Mz (Nm)Sq (µm)SskSku
1800.060.5117.00216.5084.800.590.680.495.71
21000.060.589.00260.0095.200.640.46−0.284.08
3800.10.5107.00285.00110.000.950.44−0.173.62
41000.10.5116.00310.00113.000.930.45−0.324.26
5800.061.0154.00365.0099.401.010.47−0.014.37
61000.061.0116.50315.0096.001.030.46−0.174.13
7800.11.0195.00412.50124.001.250.57−0.243.54
81000.11.0182.50435.00129.001.300.60−0.173.32
9900.080.75125.50330.00137.600.880.48−0.194.61
10900.080.75119.00335.00125.000.910.46−0.083.59
Oil + Graphitevc (m/min)f (mm/rev)ap (mm)Fc (N)Ff (N)Fp (N)Mz (Nm)Sq (µm)SskSku
1800.060.580.00192.5052.000.550.45−1.0515.30
21000.060.584.00217.5063.000.630.390.003.28
3800.10.5127.50242.5071.000.770.450.013.76
41000.10.5115.00255.0081.000.780.44−0.274.11
5800.061.0127.50375.0077.001.040.41−0.486.17
61000.061.0142.50410.0088.001.020.40−0.234.06
7800.11.0175.00485.00112.001.340.56−0.063.26
81000.11.0190.00465.00121.001.360.58−0.133.33
9900.080.75130.00410.00124.001.080.44−0.304.73
10900.080.75140.00395.00116.001.060.40−0.345.11
Table 9. Statistical functions for the prediction of dependent variables—Mathematical Function.
Table 9. Statistical functions for the prediction of dependent variables—Mathematical Function.
DryMathematicalFunction
Fc= 253.250 − 3.99675 × vc − 867.125 × f + 56.15 × f × vc + 0.655 × ap × vc 768.999 × ap × f + 201.2475
Ff= −451.195 + 6.336 × vc + 4113.75 × f − 49.990 × f × vc − 3.023 × ap × vc + 1845.0 × ap × f + 133.2
Fp= −212.050 + 2.299 × vc + 1984.197 × f − 8.112 × f × vc − 1.865 × ap × vc − 151.680 × ap × f + 148.705
Mz= −0.435 + 0.0045 × vc + 10.75 × f − 0.05 × f × vc − 0.0009 × ap × vc + 5.0 × ap × f + 0.36
Sq= −4.94 + 0.05 × vc + 39.99 × f −0.34 × f × vc − 0.028 × ap × vc − 7.4 × ap × f + 3.01
Ssk= −0.59 + 0.006 × vc + 12.475 × f − 0.185 × f × vc + 0.009 × ap × vc + 3.7 × ap × f − 0.595
Sku= 17.67 − 0.15 × vc − 93.925 × f + 1.00125 × vc × f + 0.099 × ap × vc + 1.80 × ap × f − 6.64
OilMathematical Function
Fc= 375.025 − 2.8 × vc − 4400 × f + 38.750 × f × vc − 1.550 × ap × vc + 2250.0 × ap × f + 51.75
Ff= −41.225 + 1.412 × vc − 2168.750 × f + 33.750 × f × vc − 4.800 × ap × vc + 1225.0 × ap × f + 421.
Fp= 13.775 + 0.580 × vc + 298.750 × f + 0.625 × f × vc − 0.590 × ap × vc + 365.0 × ap × f + 34.95
Mz= −0.554 + 0.0017 × vc + 12.125 × f − 0.0250 × f × vc + 0.002 × ap × vc − 3.5 × ap × f + 0.63
Sq= 3.48 − 0.025 × vc − 24.69 × f + 0.168 × f × vc + 0.012 × ap × vc + 12.625 × ap × f − 1.55
Ssk= 9.01 − 0.087 × vc − 63.04 × f + 0.541 × f × vc + 0.042 × ap × vc + 11.525 × ap × f − 3.627
Sku=20.85 − 0.15 × vc − 155.44 × f + 1.43 × f × vc + 0.03 × ap × vc + 6.60 × ap × f − 3.06
Oil + GraphiteMathematical Function
Fc= 14.96 − 0.350 × vc + 1703.125 × f − 10.312 × f × vc + 1.925 × ap × vc + 412.50 × ap × f − 69.0
Ff= −414.31 + 4.875 × vc + 3921.875 × f − 42.187 × f × vc − 1.125 × ap × vc + 1937.5 × ap × f + 270.0
Fp= −27.625 + 0.7 × vc+243.75 × f − 1.875 × f × vc − 0.05 × ap × vc + 775.0× ap × f + 6.0
Mz= −0.438 + 0.006 × vc + 2.937 × f − 0.019 × f × vc − 0.004 × ap × vc + 6.75 × ap × f + 0.66
Sq= 1.191 − 0.006 × vc − 6.425 × f + 0.036 × f × vc + 0.0036 × ap × vc + 7.25 × ap × f − 0.59475
Ssk= −12.026 + 0.117 × vc + 106.056×f − 1.03 × f × vc − 0.03 × ap × vc − 6.67 × ap × f + 2.589
Sku= 140.035 − 1.26 × vc − 1039.62 × f + 9.081 × f × vc + 0.48 × ap × vc + 176.87 × ap × f − 46.74
Table 10. Results of the application of the desirability optimization function.
Table 10. Results of the application of the desirability optimization function.
Trialvc (m/min)f (mm/rev)ap (mm)%Confidence
Dry90.00.060.554.0
Oil100.00.070.566.0
Oil + Graphite100.00.060.569.0
Table 11. Samples chosen randomly for testing the training of the ANNs created.
Table 11. Samples chosen randomly for testing the training of the ANNs created.
Dryvc (m/min)f (mm/rev)ap (mm)Fc (N)Ff (N)Fp (N)Mz (Nm)Sq(µm)SskSku
61000.061.0396.81100.5070.970.981.310.093.65
7800.11.0510.50172.62119.371.441.460.043.07
9900.080.75570.50164.42203.951.141.51−0.094.30
Oilvc (m/min)f (mm/rev)ap (mm)Fc (N)Ff (N)Fp (N)Mz (Nm)Sq(µm)SskSku
1800.060.5117.00216.5084.800.590.680.495.71
41000.10.5116.00310.00113.000.930.45−0.324.26
9900.080.75125.50330.00137.600.880.48−0.194.61
Oil + Graphitevc (m/min)f (mm/rev)ap (mm)Fc (N)Ff (N)Fp (N)Mz (Nm)Sq (µm)SskSku
21000.060.584.00217.5063.000.630.400.003.28
7800.11.0175.00485.00112.001.340.56−0.063.26
9900.080.75130.00410.00124.001.080.44−0.304.73
Table 12. Summary of active networks.
Table 12. Summary of active networks.
SummaryDryOilOil + Graphite
Index421
Network NameMLP 3-8-7MLP 3-24-7MLP 3-16-7
Training1.01.01.0
Test1.01.01.0
Validation1.01.01.0
Training error0.00.00.0
Test error0.00.00.0
Validation error0.00.00.0
Training algorithmBFGS 237BFGS 128BFGS 68
Error functionSOSSOSSOS
Hidden activationTanhTanhExponential
Output activationIdentityIdentityIdentity
Table 13. Custom predictions—Forces and torque.
Table 13. Custom predictions—Forces and torque.
DryFcFc_OutputFfFf_OutputFpFp_OutputMzMz_Output
1396.81385.77100.50130.9370.9791.130.980.98
2510.50500.35172.6295.27119.3781.321.441.36
3570.50617.46164.42207.93203.95207.951.141.31
OilFcFc_OutputFfFf_OutputFpFp_OutputMzMz_Output
1117.0081.04216.50255.5984.80102.080.590.59
2116.0090.17310.00298.07113.00123.280.930.93
3125.50119.00330.00334.99137.60124.990.880.91
Oil + GraphiteFcFc_OutputFfFf_OutputFpFp_OutputMzMz_Output
184.00102.16217.50296.7163.0090.200.630.75
2175.00199.59485.00533.07112.00154.141.341.51
3130.00139.99410.00394.99124.00115.991.081.06
Table 14. Custom predictions—Superficial Roughness.
Table 14. Custom predictions—Superficial Roughness.
DrySqSq_OutputSskSsk_OutputSkuSku_Output
11.311.460.09−0.033.652.81
21.460.930.04−0.113.073.98
31.511.57−0.090.354.305.39
OilSqSq_OutputSskSsk_OutputSkuSku_Output
10.680.370.49−0.045.714.17
20.450.46−0.32−0.124.263.33
30.480.46−0.19−0.074.613.58
Oil + GraphiteSqSq_OutputSskSsk_OutputSkuSku_Output
10.400.350.00−0.553.287.20
20.560.52−0.06−0.013.262.14
30.440.41−0.30−0.334.735.10

Share and Cite

MDPI and ACS Style

Paschoalinoto, N.W.; Batalha, G.F.; Bordinassi, E.C.; Ferrer, J.A.G.; Filho, A.F.d.L.; Ribeiro, G.d.L.X.; Cardoso, C. MQL Strategies Applied in Ti-6Al-4V Alloy Milling—Comparative Analysis between Experimental Design and Artificial Neural Networks. Materials 2020, 13, 3828. https://doi.org/10.3390/ma13173828

AMA Style

Paschoalinoto NW, Batalha GF, Bordinassi EC, Ferrer JAG, Filho AFdL, Ribeiro GdLX, Cardoso C. MQL Strategies Applied in Ti-6Al-4V Alloy Milling—Comparative Analysis between Experimental Design and Artificial Neural Networks. Materials. 2020; 13(17):3828. https://doi.org/10.3390/ma13173828

Chicago/Turabian Style

Paschoalinoto, Nelson Wilson, Gilmar Ferreira Batalha, Ed Claudio Bordinassi, Jorge Antonio Giles Ferrer, Aderval Ferreira de Lima Filho, Gleicy de L. X. Ribeiro, and Cristiano Cardoso. 2020. "MQL Strategies Applied in Ti-6Al-4V Alloy Milling—Comparative Analysis between Experimental Design and Artificial Neural Networks" Materials 13, no. 17: 3828. https://doi.org/10.3390/ma13173828

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop