Machinability of wood-plastic composites from the CNC milling process using the Box-Behnken design and response surface methodology for building applications

Abstract

This research investigated the optimization of CNC milling parameters on hardness and surface roughness properties resulting from variations in speed (220, 470, and 720 rpm), feed rate (200, 300, and 400 mm/min), and depth of cut (3, 5, and 7 mm) based on experimental design. This study aimed to evaluate the effects and relation of milling parameters using different end mill diameters (8 and 12 mm) for machining wood-plastic composites (WPCs). All of the experimental runs were determined by the Box-Behnken design and optimized using the response surface methodology. The findings from this study revealed that the main milling parameters significantly affected the hardness, average roughness (Ra), and mean peak-to-valley height (Rz) of the WPCs. Increasing the speed in a range from 220 to 720 rpm resulted in higher hardness values. On the other hand, the Ra and Rz was decreased. Additionally, this experimental result is different from the morphological structure and surface observation. It was seen that the CNC milling conditions using high speed at 720 rpm displayed smooth surfaces, which resulted in visible evenness on the WPC surfaces. Finally, numerical optimization is a good technique for the experimental results and the predicted values. The predicted conditions for the CNC milling process using end mill diameters of 8 and 12 mm included a speed of 720 rpm, a feed rate of 300 mm/min, and a depth of cut of 3 mm with the best desirability of 0.973 (97.30%). These conditions were verified in the response models and confirmed the optimal values from observed values for the variables included in the models. Also, these optimal conditions for the CNC milling parameters can be used on other types of WPCs with melting points of plastic higher than 180°C.

Keywords

Recycled polypropylene rubberwood flour response surface methodology construction materials Computer Numerical Control milling process

Introduction

Wood-plastic composites (WPCs) are one of the important products that are increasingly employed as interior and exterior building materials.^1,2 These WPC products exhibit multifunctional characteristics, such as a combination of exceptional mechanical properties and flame retardancy, which can be achieved with the addition of specific additives during composite processing.^3,4 WPCs are manufactured by thoroughly mixing ground wood fiber as reinforcement and heated thermoplastic as the matrix.^5–7 Commonly used thermoplastics include polyethylene (PE), polystyrene (PS), polypropylene (PP), polylactic acid (PLA), and others.^8,9 Among all plastic types, the use of PP has increased significantly for economic and environmental reasons in society similar to plastic recycling.^7,10,11 The use of such plastics as the matrix in WPCs is one of the growing areas of interest because of the expanding WPC market.¹² Also, cost competitiveness is achieved by taking advantage of cheap, abundant, and biodegradable wood fibers as structural fillers.^13,14 The compounding and forming processes vary based on the product types and manufacturing facility.^2,15 Generally, WPC products such as lumber or panel are produced with three main processes including injection molding, extrusion, and compression molding.^1,4,16 Additionally, the machining process is an important part of the manufacturing of WPC materials. This process is considered to give the WPC products the desired shape. The requirements for higher machining quality and manufacturing efficiency of materials have led to the application of WPC products that have a variety of uses.^3,17

The milling process is one of the most significant manufacturing processes in WPCs-cutting operations. This manufacturing process is planned and improved to obtain either high quality or low cost.^18,19 In the manufacturing industry, new techniques have been developed for the manufacturing performance through better process productivity, improved part quality, and advanced tool performance.²⁰ To keep increasing the mill machining performance, the computer numerical control (CNC) technique has been recently developed to replace the conventional machining process.^21,22 The CNC method is a manufacturing machine technology that is widely used in many industries. The CNC machine is used to manufacture high-precision mass products. This technique employs a production process that automates the movement, control, and precision of the machine and cutting tools through the use of preprogrammed computer software.^18,23 CNC is widely used in manufacturing for machining metal, plastic parts, and composites. Milling is a common cutting process and the operation can also be automated with CNC.^24,25 Additionally, it has been found that the tool geometries and certain cutting parameters also influence these effects. The effects of cutting speed, feed rate, and depth of cut including surface quality on the machinability of WPCs were assessed.^19,21,23 Guo et al.²⁶ reported that the changes in cutting forces resulted in surface roughness quality during orthogonal machining of WPCs, and they were positively related to cutting depth. Their results also confirmed that the process resulted in smooth roughness quality when the cutting depth was low. Pei et al.²⁷ affirmed that the cutting depth contributed the most to cutting temperature, followed by spindle speed and feed rate. This behavior shows that a series of carefully designed milling experiments and surface defects on WPCs are strongly linked to milling factors and material properties.^18,22

The Box-Behnken design (BBD) was used to determine the experimental runs of the CNC milling process for the WPCs. However, the BBD is still considered to be more proficient and powerful than other designs such as the three-level full factorial design, central composite design (CCD), and Doehlert design.^20,24 In general, the factors are set at one of three equally spaced values, usually coded as low (−1), medium (0), and high (+1).^24,28 Also, this process is the experimental design for response surface methodology (RSM). The RSM, using a proper design of experiment, has been used extensively.^1,3,21,29 The RSM is a collection of statistical analyses useful for the development, improvement, and optimization processes applicable in almost every field.^1,23,25 It is worth applying the conditions when numerous variables (independent variables) potentially influence the quality characteristics (response variables) of the product process.³⁰ The basic idea of RSM is to use a sequence of designed experiments to obtain an optimal response.

In this study, the requirements for higher machining quality and manufacturing efficiency led to a significant amount of research aimed at control and monitoring. It is not surprising that many researchers have focused on understanding these aspects of WPC machining. Thus, the objective of the present work was to investigate the effects of the CNC milling parameters on hardness and surface roughness responses during the machining of WPC materials and to optimize these milling parameters using the RSM. Furthermore, the surface changes and interfacial morphologies of the WPCs after the milling process were also observed using optical microscopy and Field Emission Scanning Electron Microscopy (FE-SEM).

Materials and methods

Materials

The plastic matrix used in this study was acquired from Southern Plastic Co., Ltd (Songkhla, Thailand) and comprised recycled polypropylene (rPP) with a density of 0.93 g/cm³ and a melt flow index of 11 g/10 min at 230°C. The rubberwood flour (RWF) used as a reinforcing material was supplied by a local furniture company (Trang, Thailand). Before compounding, the RWF was screened on a sieve to classify the particle size of 80 mesh (less than 177 microns) and then dried in an oven at a temperature of 110°C for 8 h to remove the moisture content (less than 1%). The calcium carbonate (CC) modified surface, used as a filler in the WPCs, was purchased from Sand and Soy Industry Company Ltd, (Lopburi, Thailand). The coupling agent in this study was maleic-anhydride-grafted polypropylene (MAPP) with a grafting ratio of 8–10% maleic anhydride, which was provided by Sigma-Aldrich (Missouri, USA). It was used as the interfacial bonding between matrix and reinforcement. The ultraviolet stabilizer (UV stabilizer) with MEUV008 coding was purchased from TH Color Company Ltd (Samut Prakan, Thailand). The paraffin wax as a lubricant (Lub) was also supplied by Nippon Seiro Company Ltd (Yamaguchi, Japan).

Preparation of the wood-plastic composite samples

The optimal formulation for construction material used in this research was 51.8 wt% rPP, 35.9 wt% RWF, 7.2 wt% CC, 3.9 wt% MAPP, 0.2 wt% UV stabilizer, and 1.0 wt% Lub.³¹ Initially, these components were mixed and melt-blended into WPC materials using a twin-screw extruder (Model CTE-D25L40 from Chareon Tut Co., Ltd, Samut Prakan, Thailand). The seven temperature zones from the hopper to the die were 170, 175, 180, 185, 185, 190, and 200°C, to reduce the degradation of the WPC compositions.³² The screw rotation speed was set to 40°rpm and a pressure of 6.89 MPa. The extruder strand passed through a cutting pellet machine with WPC pellet sizes of 2-3 mm. Prior to compression, the WPC pellets were dried in an oven at a temperature of 50°C for 8 h to remove moisture content (less than 1%). Subsequently, the WPC pellets were weighed at 200 g and then molded using a compression molding machine from a rectangular mold with dimensions of 50 mm (width) × 50 mm (length) × 30 mm (thickness). The best conditions for the compressing process were set at a temperature of 190°C under a pressure of 10.34 MPa for 20 min with processes of pre-heating (5 min), compressing (10 min), and cooling (5 min), respectively. Experiment was carried out using 90 samples.

Experimental design and optimization of CNC milling conditions

The BBD was used to determine the CNC milling conditions against three independent variables. All of the independent variables for this research were at different levels of speed (A), feed rate (B), and depth of cut (C) which affected the surface quality of the WPCs. Generally, the independent variable was placed at one of three equally spaced values, usually coded as low (−1), medium (0), and high (+1). The actual levels and codes of experimental design are displayed in Table 1. The inclusion of 17 experimental runs, as suggested using Design-Expert software (version 8.0.6, Stat Inc.), was used to analyze the experimental data to allow for a better estimate of the experimental error and to provide an extra information region. The response variables included hardness value (Y₁), average roughness (Y₂), and mean peak-to-valley height (Y₃). The experimental runs and response variables are illustrated in Table 2.

Table 1.

The CNC milling parameters and Box-Behnken design of experiments.

Factor	Code	Parameter level
Factor	Code	Low (−)	Meduim (0)	High (+)
Speed (rpm)	A	220	470	720
Feed rate (mm/min)	B	200	300	400
Depth of cut (mm)	C	3	5	7

Table 2.

Experimental runs using Box-Behnken design and responses.

Run	Milling parameter			Hardness property (shore D)		Surface roughness property (µm)
Run	Arpm	Bmm/min	Cmm	End mill of 8 mm	End mill of 12 mm	Ra in end mill of 8 mm	Ra in end mill of 12 mm	Rz in end mill of 8 mm	Rz in end mill of 12 mm
1	470	200	3	77.23	78.20	0.79	1.15	8.79	13.85
2	470	300	5	75.23	77.25	0.97	1.13	13.72	11.56
3^a	470	300	5	76.22	77.56	0.90	1.42	11.28	13.60
4	220	200	5	74.12	75.86	1.29	1.66	12.03	12.81
5	220	300	3	70.28	71.56	2.30	2.05	14.75	14.62
6	720	300	3	75.20	77.15	0.96	1.49	9.93	15.66
7	720	400	5	78.65	78.96	0.77	1.24	9.72	12.50
8^a	470*	300	5	77.12	78.25	0.79	1.15	11.76	12.21
9	470	400	7	75.44	76.12	0.96	1.13	8.71	10.04
10	720	200	5	78.89	79.65	0.78	0.94	12.21	13.43
11	220	300	7	70.29	72.01	1.91	1.02	11.32	11.02
12	220	400	5	73.55	75.20	1.26	1.43	12.22	10.93
13	470	400	3	70.15	72.02	1.79	1.61	13.27	14.94
14	470	200	7	72.56	73.22	1.57	1.38	11.64	10.63
15^a	470	300	5	71.55	72.56	1.65	1.68	14.78	12.41
16	470	300	3	74.33	75.89	1.23	0.99	12.45	12.44
17	720	300	7	73.22	74.08	1.42	1.32	10.84	11.90

Note: ^a Duplicate experiments.

The data for each CNC milling condition following the experimental design were analyzed using ANOVA with a 5% significance level (α = 0.05) to identify the significant effects of each factor. Multiple regression coefficients were used to analyze the relationships of several variables exhibited in linear, quadratic, and interaction terms of independent variables in response surface models. The general regression models were assumed to relate to these responses, as exhibited in equation (1). Also, the response surface methodology (RSM) was used to optimize the quality parameters of CNC mill machining. The optimal condition was presented under the scope of determined factors, which were generated in the form of response contour plots.

Y_{p r e d i c t e d} = β_{0} + \sum_{i = 1}^{k} β_{i} X_{i} + \sum_{i = 1}^{k} β_{i i} X_{i}^{2} + \sum_{i} \sum_{< j = 2}^{k} β_{i j} X_{i} X_{j} + ε_{i}

(1)

Where Y_predicted is the response parameter, k is the number of factors, β₀ is the constant coefficient, x_i, and x_j are the factors (independent variables), β_i is the linear coefficient, β_ii is the quadratic coefficient, β_ij is the interaction coefficients between the input factors x_i and x_j, and ε_i is the residual associated with the experiment (Petdee et al., 2023).

Machine setup and methodology

The WPC samples were machine-milled using a CNC machine (Model GE Fanuc Series O, EMCO Ltd, Michigan, USA). The cutting tools used in the experiment were end mills (4 flutes) with diameters of 10 and 12 mm (Model S.A.P from Smart Tech Tooling Co., Ltd, Chonburi, Thailand). The specifications of the cutting tool for the CNC milling process are listed in Table 3. The WPC specimens in this study were molded from a compression molding machine. A rectangular specimen with dimensions of 50 mm (width) × 50 mm (length) × 40 mm (thickness) is displayed in Figure 1(a). Workpieces after the CNC milling process are shown in Figure 1(b). Also, the samples were fixed on the working plane of the machine using the fixture. The CNC machine setup and equipment for the milling process showing the milling position are illustrated in Figure 2(a) and (b).

Table 3.

The specifications and details of cutting tool for CNC milling process.

Tool specification	End mill of 8 mm	End mill of 12 mm
Height of thread span (D₁)	8 mm	12 mm
Grip size (D₂)	8 mm	12 mm
End mill length (L₁)	60 mm	75 mm
Length of thread span (L₂)	20 mm	25 mm
Strength of workpiece	45 HRC	45 HRC
Milling pattern	Fine milling	Fine milling
Degree of thread range	35°	35°

Figure 1.

WPC sample used in the experiment (a) before the CNC milling process and (b) after the CNC milling process.

Figure 2.

Machine setup and equipment for the milling process showing (a) the CNC milling machine and (b) equipment setup for the milling process.

Characterizations of WPC samples

Hardness measurement

The hardness test after CNC mill machining evaluated the hardness on the surface of WPC samples. The specimens were prepared in accordance with the ASTM D2240 specification. All of the samples had approximate dimensions of 50 mm (length) × 50 mm (width) × 40 mm (thickness) and were tested employing a mechanical Shore D Durometer (Model GS-702G from Teclock Corporation, Nagano, Japan). The hardness response values were tested at room temperature (25°C) and reported the average value of five replications.

Surface roughness measurement

The surface roughness characteristics of WPCs after the CNC milling process were used to evaluate the smooth machined surface using a surface profilometer on a Portable Surface Roughness Tester (Model SJ-210 from Mitutoyo Co., Ltd. Bangkok, Thailand) with a cut-off length of 0.8 mm for the surface roughness measurements. The surface roughness parameters, namely average roughness (Ra) and mean peak-to-valley height (Rz), were considered to evaluate the quality of the surface of WPC materials. The Ra is the arithmetic mean of the absolute values for the profile deviations from the mean line and is commonly used as a parameter in finished surface measurement. Similarly, the Rz is the average of the successive values of maximum heights within a sampling length calculated over the evaluation length. The surface roughness response was presented from the digital information. The experimental data obtained from the roughness tests was processed using Design-Expert software (version 8.0.6, Stat Inc.) to calculate the Ra and Rz variables. Rectangular specimens with dimensions of 50 mm (width) × 50 mm (length) × 40 mm (thickness) were used for measuring roughness. The test was characterized at room temperature (25°C) with five replications.

Visual surface and morphological observation

The changes in the WPC surfaces with different milling parameter conditions were measured using optical microscopy (Model Zeiss Axioskop, Oberkochen, Germany). The interfacial morphology of WPCs after CNC mill machining was observed. Additionally, composite sample surfaces including voids, cracks, interfacial bonding, and the dispersion of the wood flour in the recycled plastic matrix were investigated from a cross-section area using an FE-SEM Model: FEI Apreo microscope from FEI Company (Oregon, USA). Before the FE-SEM observation, the WPC surfaces were gold-coated using a sputter coater to prevent electron charging during the imaging analysis with an accelerating voltage of 20 kV. Micrograph images were taken at a magnification level of 100x.

Results and discussion

Model adequacy checking

The model adequacy checking of hardness properties is used as an example to analyze the data normality of the CNC milling process with an end mill of 8 mm. The estimation of milling parameters testing from the hypothesis of the estimator is based on major assumptions of the experimental design.^1,23 In this study, the adequacy of the response models was investigated with various diagnostic plots, such as normal probability plots of the residuals, plots of residuals versus predicted response, and plots of predicted response versus actual values. The model adequacy of hardness properties machined using an end mill of 8 mm is presented as an example in Figure 3. According to Figure 3(a), this plot displays the normal probability plot of residuals. These data follow a normal distribution with mean and variance in which the data plots should be approximately linear. It is reasonable to assume that the error terms are normally distributed; the fitted model is the linear regression model that is normally distributed. In addition, Figure 3(b) illustrates the relationship between internally studentized residuals and the predicted response obtained from the fitted model. This stage is a visual check of the assumption of constant variance.^28,30 It is observed that the plot is randomly scattered with a consistent top-to-bottom range of residuals across the predictions on the horizontal axis. On the other hand, the response model based on the multiple regression model is used instead if the predictor on the horizontal axis is a new and different predictor. Moreover, the plot collections around the diagonal line for predicted response and actual values are shown in Figure 3(c). The continuous response values on the vertical axis are predicted and observed from a random model against the actual value.^3,31 This might explain the displayed plot collections around the diagonal line for predicted response and actual values, which suggests that the predicted responses were relatively close to the actual results. Also, the data are distributed near the straight line at an acceptable level. It can be concluded that the predicted response illustrated adequate agreement with the actual value.

Figure 3.

Model adequacy check for hardness response (end mill of 8 mm) showing (a) normal probability plot of the residuals, (b) plots of residuals versus predicted response, and (c) plots of predicted response versus actual values.

Fitting model and statistical analysis

Analysis of variance (ANOVA) is a statistical method used extensively to analyze the differences between the means of two or more groups.^3,9 It is often used to determine the statistically significant differences based on the F-test or p-value between the means of different groups.^17,23 In this research, the optimization of milling parameters based on hardness and surface roughness properties resulting from parameter variations such as speed (A), feed rate (B), and depth of cut (C) was performed using the Box-Behnken experimental design. The best-fitting model summary for Ra machined with an end mill of 8 mm with the quadratic model is exhibited as an example in Table 4. The findings from the fitted model revealed that the sequential model sum of the square for the quadratic model with a p-value of 0.0012 is significant (p < .05) based on the fitted model of Ra. The lack-of-fit in this work with a p-value of 0.0797 was insignificant (p > .05) for the quadratic model. Additionally, the adjusted R² (Adj-R²) and predicted R² (Pred-R²) were found to have 0.9012 and 0.8910, respectively. This suggested that the model performed well, as indicated by the high values for correctness and accuracy. Overall, the analysis of the alternative types of response models is concluded in Table 5. It was found that the hardness responses were fitted in the linear models, whereas the surface roughness responses were fitted in the quadratic models.

Table 4.

Fitted model summary for hardness response of WPCs machined with end mill of 8 mm.

Source	Sequential p-value	Lack of fit p-value	Adjusted	Predicted	Remarks
Source	Sequential p-value	Lack of fit p-value	R ²	R ²	Remarks
Linear	0.0002*	0.0158*	0.4512	0.7541
2FI	0.1254	0.0856	0.8451	0.8421
Quadratic	0.0012*	0.0797	0.9012	0.8910	Suggested
Cubic	0.1258	0.0845	0.8801	0.5689

Note: *p-value less than 0.05 is considered significance.

Table 5.

Analysis of variance based on the fitted linear and quadratic response models for hardness, Ra, and Rz properties.

Source	Linear response model		Quadratic response model
	Hardness		Ra		Rz
	End mill of 8 mm	End mill of 12 mm	End mill of 8 mm	End mill of 12 mm	End mill of 8 mm	End mill of 12 mm
Model	0.0001*	0.0002*	0.0024*	0.0019*	0.0002*	<0.0001*
A-Speed	0.0102*	0.0004*	0.0145*	0.0022*	0.0004*	<0.0001*
B-Feed Rate	0.2145	0.1259	0.2524	0.0894	0.1089	0.0698
C-Depth of Cut	<0.0001*	<0.0001*	0.0081*	0.0014*	0.0147*	<0.0001*
AB	-	-	0.0004*	<0.0001*	0.0001*	0.0001*
AC	-	-	0.1141	0.1270	0.1217	0.1102
BC	-	-	0.0292*	0.0304*	0.0145*	0.0498*
Lack of fit	0.0797	0.1245	0.0844	0.1124	0.0841	0.0855
R ²	0.9145 (91.45%)	0.8921 (89.21%)	0.8891 (88.91%)	0.9045 (90.45%)	0.9215 (92.15%)	0.8795 (87.95%)
Adj-R²	0.9012 (90.12%)	0.8752 (87.52%)	0.8621 (86.21%)	0.8856 (88.56%)	0.9001 (90.01%)	0.8612 (86.12%)
Pred-R²	0.8910 (89.10%)	0.8622 (86.22%)	0.8541 (85.41%)	0.8601 (86.01%)	0.8979 (89.79%)	0.8498 (84.98%)
C.V.	4.58%	2.89%	7.12%	7.31%	8.03%	4.01%

Note: *p-value less than 0.05 is considered significance.

The ANOVA analysis for both linear and quadratic models from the experimental results was also studied. It can be seen that all of the response models, namely hardness, Ra, and Rz properties, exhibited the acceptable values and statistical significance (p < .05) of these models. The multiple interactions of the main factors affected Rz of the WPCs after the CNC milling process with an end mill of 8 mm, namely speed (A), feed rate (B), and depth of cut (C), are shown in Figure 4. Additionally, the interaction terms of response models are also displayed in Table 5. The response models show significant interactions between speed (A) and feed rate (B), and between feed rate (B) and depth of cut (C) for Ra and Rz responses. This is in agreement with the change rule of the model responses with the input parameters examined in previous studies.^18,24

Figure 4.

Main parameters affecting Rz of the CNC milling process with an end mill of 8 mm on (a) speed, (b) feed rate, and (c) depth of cut.

Finally, the ANOVA analysis also displayed statistical values such as lack-of-fit, coefficients of determination (R²), Adj-R², Pred-R², and coefficients of variation (C.V.). All of the p-values and data analysis are summarized in Table 5. It was observed that R² values ranging from 0.8795 (87.95%) to 0.9215 (92.15%) were found in the response models for the milling process. The analyzed data suggested both the good fit and reliability of the experimental values.^12,31 The response of Rz was used as an example to analyze the model of the CNC milling process with an end mill of 8 mm. The Adj-R² value was 0.9001 (90.01%), which displayed the adequacy of WPC samples in the experimental design. The Pred-R² value of the Rz was 0.8979, indicating the model accounts for 89.79 % of the variability in the new terms. These results could be explained with the regression model analysis to indicate the predicted responses for new observations.^9,23 Also, the percentage of C.V. was used to analyze the relative dispersion of the residual variation from the predicted response model.^3,14 The C.V. value of Rz was 8.03%, which indicates the best precision and fit for the parametric models. The C.V. value should not be greater than 10 %, in which high C.V. shows high data variation in the mean value and does not lead to the satisfactory development of adequate response models.^1,3,23 Therefore, the low C.V. value displays the best precision and fit for parametric models.

Effect of CNC milling parameters on hardness property and optimization

This research presents the optimization of CNC milling parameters on hardness properties resulting from variations in speed, feed rate, and depth of cut based on experimental design. The response values in Table 2 display the hardness values after machining with different end mill diameters of 8 and 12 mm. Also, the hardness values were changed with different parameters in the CNC milling process.^23,25 The results showed that the hardness values after mill machining were found in ranges from 70.15 to 78.89 for an end mill of 8 mm and from 71.56 to 79.65 for an end mill of 12 mm. The highest hardness values of 78.89 and 79.65 were found at a speed of 720 rpm, a feed rate of 200 mm/min, and a depth of cut of 5 mm. In addition, the fitted regression equations in terms of parameters to predict the effects of conditions on hardness values are expressed by the linear models, as in equations (2) and (3) below:

Hardness (end mill of 8 mm) = 3.19 - 26.08 A - 1.47 B - 17.34 C

(2)

Hardness (end mill of 12 mm) = 4.28 - 32.18 A - 2.72 B - 14.85 C

(3)

As seen in both equations, the speed (A) of CNC milling parameters has the largest negative coefficients in the fitted models at 26.08 and 32.18 for end mills of 8 mm and 12 mm, respectively. It means that the hardness values were improved after the change of speed parameter for the milling condition. The increase of speed value at 720 rpm resulted in high heat and density values on the WPC surface after the milling process. This leads to the increased strength of the surface material compared to those at speed values of 220 and 470 rpm.³³ Similarly, it can be observed that the depth of cut (C) showed a high negative coefficient in the model at end mills of 8 mm and 12 mm for hardness values of 17.34 and 14.85, respectively. The above results confirm that an increase in the depth-of-cut of the CNC milling process provided poor surface strength as evaluated from the hardness property.

Tables 6(a) and (b) show the cross-section images obtained using the FE-SEM observation. It can be seen that the CNC milling condition using a high speed of 720 rpm displayed a smooth surface, which resulted in visible evenness on the WPC surface. This result is different from the FE-SEM morphology. Another notable observation is the fact that the smoothness of WPC surfaces increased the hardness values. This may be a result of WPC material deformation occurring and a part of the chip being melted and added to the surfaces. The composites are improved to enhance intrinsic adhesion between WPC surfaces and chips coating, providing stronger interfacial area and a mechanical interlocking mechanism.^15,26

Table 6.

SEM morphological analysis and optical microscopy observation of WPCs with different speed parameters.

Speed	220 rpm	720 rpm
Feed rate: 300 mm/min
Depth of cut: 3 mm
Feed rate: 300 mm/min
Depth of cut: 3 mm

The predicted CNC milling condition from three main parameters, using end mill diameters of 8 and 12 mm based on hardness responses, is displayed in Table 7. The optimal conditions for hardness properties with end mills of both 8 mm and 12 mm included a speed (A) of 720 rpm, a feed rate (B) of 220 mm/min, and a depth of cut (C) of 5 mm with a combined high-desirability score value of 1.000 (100%). In light of the above research, the optimal parameters for hardness properties agreed with the research of Homkhiew et al.,³ leading to the optimization of several applications that depend on their prioritization. Also, this CNC milling condition of the WPC materials can be used for the most suitable applications based on the hardness properties.

Table 7.

Predicted responses and desirability score with optimal condition for all of the properties.

Response	Optimized CNC milling factor			Predicted response	Desirability
Response	Speed (A)	Feed rate (B)	Depth of cut (C)	Predicted response	Desirability
Hardness (end mill 8 mm)	720	220	5	79.23	1.000 (100%)
Hardness (end mill 12 mm)	720	220	5	78.23	1.000 (100%)
Ra (end mill 8 mm)	720	220	3	0.77	0.971 (97.1%)
Ra (end mill 12 mm)	720	220	3	0.97	0.984 (98.4%)
Rz (end mill 8 mm)	720	400	3	8.74	0.997 (99.7%)
Rz (end mill 12 mm)	720	400	3	10.52	0.963 (96.3%)

Effects of CNC milling parameters on Ra properties and optimization

The average surface roughness (Ra) of the composite samples after the CNC milling process is summarized in Table 2. The effect of the examined milling parameters showed the interactions and second-order on the Ra as a three-dimensional (3D) image based on the response surface. The Ra values of the WPC samples were affected significantly by different CNC milling conditions. The significant differences for milling parameters were determined individually for these tests by ANOVA and optimized conditions using RSM.^1,3,29,31 Additionally, the fitted quadratic regression models with the observed values of the Ra response are shown in equations (4) and (5).

R a (end mill of 8 mm) = 5.12 + 3.47 A + 0.89 B + 2.59 C - 0.42 A B - 0.02 A C - 0.33 B C

33.45 A^{2} + 0.0025 B^{2} + 15.48 C^{2}

(4)

R a (end mill of 12 mm) = 6.08 + 5.82 A + 1.42 B + 4.86 C - 0.72 A B - 0.08 A C - 0.39 B C

45.08 A^{2} + 0.0049 B^{2} + 23.82 C^{2}

(5)

The relation between average Ra values and three examined parameters was analyzed using quadratic models. The Ra equations at both cutting tool diameters display the positive coefficients for all the main parameters, whereas the interaction terms suggest negative coefficients. It is observed that the speed (A) gave the highest positive coefficient values of 3.47 and 5.82 in the quadratic models for Ra values at both end mills of 8 mm and 12 mm, respectively. The Ra value of the WPCs constantly improved with an increase of the speed parameters in a range from 220 to 720 rpm. This result of the experiment agrees with Slamani et al.,³⁵ who reported that increasing the high spindle speed parameter in a range from 200 to 700 rpm achieved better quality composite surfaces.

The 3D response surface and contour plots of Ra after machining with an end mill of 8 mm are exhibited in Figure 5. The RSM analysis showed some significant differences among the CNC milling conditions for the Ra values.^24,25,36 Figure 5(a) shows the effects of interaction between speed and feed rate parameters after the milling process using an end mill diameter of 8 mm. The outcome of this study indicated that a CNC milling speed ranging from 220 to 720 rpm affected the Ra response of WPC samples. The Ra values of the WPC surfaces curvedly decreased from 15.0 to 10.0 µm with the increase in the speed parameter. On the other hand, the increasing feed rate ranged from 200 to 300 mm/min and slowly increased the Ra value, after that, it decreased gradually. The interaction parameters between speed and depth of cut are displayed in Figure 5(b). The findings from this study revealed that the Ra values of the WPCs enhanced to the maximum values with an increment of the depth of cut ranging from 3 to 7 mm. This result of the experiment is in line with Doluk et al.,³⁷ who reported that an increment in the depth of cut for the mill machining of the WPCs slightly affected the Ra values, which increased from 2.5 to 2.9 µm. Figure 5(c) indicates the 3D response and interaction contour plots between feed rate and depth of cut. However, the Ra values of the WPCs rose with an increasing feed rate in a range from 200 to 400 mm/min. The reason for this phenomenon is the high temperature in the cutting zone due to higher friction between the cutting edge tip and WPC samples. In addition, an increase in the depth of cut in a range from 3 to 7 mm slowly enhanced the Ra values of the WPCs. This condition leads to poor surface performance.

Figure 5.

Effects of CNC milling parameters with an end mill of 8 mm and response surface contour plots on Ra showing (a) speed and feed rate (b) speed and depth of cut, and (c) feed rate and depth of cut.

Similarly, the response surface optimization and contour plots of Ra after machining with an end mill of 12 mm are also displayed in Figure 6. The experimental results are similar to those of the milling process using an end mill diameter of 8 mm. The minimum Ra value of 1.10 µm was found in the speed parameter at 720 rpm, as presented in Figure 6(a). In addition, the WPC surface performance with a feed rate of 200 mm/min had lower Ra values than those of 300 and 400 mm/min. This may be a result of the slower axial movement of the end mill, resulting less material deformation and the complete removal of chips from the required WPC surfaces.^23,38 Furthermore, the relation between speed and depth of cut based on response surface plots is illustrated in Figure 6(b). It can be seen that an increment of the speed in a range of 620 to 720 rpm decreased the Ra response of the composite samples from 1.3 to 1.2 µm. This is due to the cutting amount per unit time being reduced with the increasing speed parameter, and the cutting resistance force decreased.^18,26,35 Moreover, the effects of both feed rate and depth of cut for the CNC milling process on Ra properties are exhibited in Figure 6(c). Machining with a feed rate ranging from 200 to 300 mm/min and a depth of cut ranging from 3 to 4 mm exhibited less roughness surface, more smoothness, and great interfacial adhesion.

Figure 6.

Effects of CNC milling parameters with an end mill of 12 mm and response surface contour plots on Ra showing (a) speed and feed rate (b) speed and depth of cut, and (c) feed rate and depth of cut.

Tables 6(c) and (d) show morphological images of WPC surface observation obtained from the optical microscopy. The WPC material deformation occurs in the area around mill machining when a part of the chip is melted and added to the WPC surface. Also, the chips of WPC samples sticking to the adjacent back part of the cutting tools occurred during the process of machining. Comparing the optical microscopy images with speeds of 220 and 720 rpm on Ra properties, the composites machined with a speed of 720 rpm had smoother surfaces than those of 220 rpm. As mentioned above, this condition enabled better mill machining quality.

The optimal conditions of the CNC milling process from main parameters, namely speed, feed rate, and depth of cut, using end mill diameters of 8 and 12 mm based on Ra properties are depicted in Table 6. The optimal conditions for the Ra value with end mills of both 8 mm and 12 mm included a speed (A) of 720 rpm, a feed rate (B) of 220 mm/min, and a depth of cut (C) of 3 mm. The high-desirability scores based on predicted Ra response were 0.971 (97.10%) and 0.984 (98.40%) for end mills of 8 mm and 12 mm, respectively. As expected, these optimal conditions of the WPC samples can be used for the most suitable applications in the CNC milling process based on the Ra properties.

Effects of CNC milling parameters on Rz property and optimization

The mean peak-to-valley height (Rz) was considered to evaluate the efficiency on the surface of WPC samples. The speed (A), feed rate (B), and depth of cut (C) are major parameters that play significant roles in the Rz values of WPC materials.^18,35 The CNC milling process of the different conditions on the surface of WPCs establishes the smooth surface quality of the product. The experimental results of Rz response after the milling process using end mills of both 8 mm and 12 mm are shown in Table 2. The effect of the WPC surfaces at the minimum Rz is statistically significant. These results were evaluated by ANOVA and optimized conditions using the RSM.^1,9,23,24 Consequently, the fitted regression equations in terms of actual factors of the Rz responses are displayed in equations (6) and (7).

R z (end mill of 8 mm) = 8.45 + 5.61 A + 0.72 B + 4.77 C - 0.75 A B - 0.06 A C - 0.51 B C

51.48 A^{2} + 0.0019 B^{2} + 46.71 C^{2}

(6)

R z (end mill of 12 mm) = 9.12 + 9.01 A + 1.43 B + 7.64 C - 0.82 A B - 0.08 A C - 0.64 B C

67.58 A^{2} + 0.0074 B^{2} + 39.18 C^{2}

(7)

It should be noted that the response models for Rz at all of the main parameters (speed (A), feed rate (B), and depth of cut (C)) only display positive coefficients. The parameters were constructed using values calculated as the arithmetic average of Rz values measured on the WPC surfaces. It is observed that the highest positive coefficient values of 3.47 and 5.82 were found in the speed parameter at both end mills of 8 mm and 12 mm, which is similar to the Ra response. In addition, the interaction terms of quadratic models for Rz responses show negative coefficients. The highest negative coefficient values of 0.75 and 0.82 were found in interaction terms between the speed (A) and feed rate (B) of Rz equations.

The response surface plots of Rz after the CNC milling process with an end mill of 8 mm are exhibited in Figure 7. The average values present the effects of multi-relation between the speed and feed rate parameters on the Rz property. Figure 7(a) exhibits a visible change in the course of Rz depending on the increase in feed rate. The Rz values rise with increasing revolution at feed rates of 200 and 400 mm/min, whereas revolution from 220 to 720 rpm displays values ranging in a narrow interval from less than 11.0 up to 13.0 ?m. As reported by Rabiei et al.,²⁵ there is a relationship between the cutting speed and feed rate of the surface quality for WPC materials of three samples with different CNC milling conditions. In addition, Figure 7(b) shows that the milling parameters, namely speed between 620 and 720 rpm and depth of cut between 3 and 4 mm, presented the lowest values in Rz, which decreased in a range from 14.0 to 11.0 ?m compared to the other conditions. Also, the interaction parameters based on feed rate and depth of cut for Rz response are shown in Figure 7(c). It can be seen from the contour plots that the feed rate in a range from 200 to 300 mm/min had the minimum Rz for the milling process using an end mill diameter of 8 mm, whereas a depth of cut at 3 mm showed better milling surface quality when compared to those of 5 and 7 mm for WPCs machining. This result agrees with Geier and Pereszlai,³⁹ who reported that decreasing the depth of cut in the mill machining of the WPCs decreased the Rz values. This is because the deformation resistance and turning force of the total area of the milling decreased.

Figure 7.

Effects of CNC milling parameters with an end mill of 8 mm and response surface contour plots on Rz showing (a) speed and feed rate (b) speed and depth of cut, and (c) feed rate and depth of cut.

The response contour plots of Rz after CNC mill machining with an end mill of 12 mm are presented in Figure 8. Figure 8(a) indicates that an increment of the speed from 220 to 720 rpm in the milling process slightly increased the Rz from 12.0 to 13.0 µm. Additionally, Rz values increased with an increase in the depth of cut in a range from 3 to 7 mm, as displayed in Figure 8(b). This can be explained by the fact that the total area of the milling enhances as the milling depth rises and the machining force increases.^35,40 Likewise, increasing the feed rate in a range from 200 to 400 mm/min increased the Rz of the WPCs from 11.0 to 14.0 µm, as shown in Figure 8(c). This may be a result of an increase in distance between the successive grooves due to faster axial movement of the cutting tool on the WPCs workpiece, and the chips are partially removed from the surface.

Figure 8.

Effects of CNC milling parameters with an end mill of 12 mm and response surface contour plots on Rz showing (a) speed and feed rate (b) speed and depth of cut, and (c) feed rate and depth of cut.

The optical microscopy observation of the WPC material surfaces displayed the heterogeneity of the composites and also the presence of microcracks located on contacts between wood fiber and the plastic matrix. As seen in Table 6, the WPC surface observation is in agreement with Ra properties. It was found that the results of speed comparison (220 and 720 rpm) in this study showed that the speed parameter of 720 rpm had a lower Rz value than the 220 rpm. The results of the comparison in this study implied that this speed parameter condition was the optimal condition for the CNC milling process based on the Rz property.

The result is consistent with the predicted value. This makes it possible to predict the course of roughness parameters for variable parameters (depending on the conditions of the milling process). The predicted conditions of Rz response after the CNC milling process with end mills of 8 and 12 mm are displayed in Table 7. The optimal conditions for Rz response with end mills of 8 and 12 mm include a speed (A) of 720 rpm, a feed rate (B) of 400 mm/min, and a depth of cut (C) of 3 mm. The high-desirability scores for Rz were in agreement with the values of 0.997 (99.70%) and 0.963 (96.30%). These optimal conditions of the WPCs represent the most suitable parameters for the CNC milling process based on the Rz response.

Optimization of CNC milling parameters for the WPCs

The numerical optimization technique with multiple responses based on the desirability approach is usually conducted using Design-Expert software (Version 8.0.6). The input parameters, namely speed, feed rate, and depth of cut, were chosen in a range of experimental designs to optimize the milling process conditions. The goals of the output characteristics were selected using the highest hardness value and the lowest Ra and Rz values. The milling conditions of all the properties were optimized using the RSM. The optimized condition for the CNC milling parameters along with the best desirability score are illustrated in Table 8. The outcome of this study stated that the optimal condition from overall properties was a speed (A) of 720 rpm, a feed rate (B) of 300 mm/min, and a depth of cut (C) of 3 mm, which provides the high desirability score of 0.973. It is suggested that the mathematical model can predict the response to an adequate extent of 97.30%. Furthermore, there is good agreement between the predicted values and observed values, verifying the validity of the response models. The predicted condition and observed values are also included in Table 8. It can be seen that the bias of those methods is less than 5%, which is an acceptable criterion. This result confirms that the optimal condition in the CNC milling process of WPCs obtained from the model was reliable.

Table 8.

Optimal condition and comparison of predicted responses and observed values for overall properties.

Response	Optimized CNC milling factor			Predicted response	Observed value
Response	Speed (A)	Feed rate (B)	Depth of cut (C)	Predicted response	Observed value
Hardness (end mill 8 mm)	720	300	3	78.85	78.02
Hardness (end mill 12 mm)				77.42	76.58
Ra (end mill 8 mm)				0.81	0.89
Ra (end mill 12 mm)				0.98	0.96
Rz (end mill 8 mm)				9.98	10.12
Rz (end mill 12 mm)				10.79	11.01

It is necessary to note that the experimental results in this work are valid only for the presented technological conditions, parameters, and cutting tools. Thus, it is recommended to study other factors in the future, such as temperature, chips, and tool wear of mill machining to determine the most suitable conditions to achieve a WPC surface with the highest quality.^34,41

Conclusions

The goal of the optimization for CNC milling parameters is to determine the minimum surface roughness and high hardness required to produce WPC products for building applications. The milling parameters, namely speed (A), feed rate (B), and depth of cut (C), are the main variables that play significant roles in the properties of WPCs. The numerical conditions were considered using the Box-Behnken design to determine the experimental runs for the CNC milling process. All of the experimental results, including the fitted regression models, were analyzed by ANOVA and optimized using the response surface methodology. These results could be explained with the following conclusions:

• All of the parameters (speed, feed rate, and depth of cut) for the CNC milling process, using end mill diameters of 8 mm and 12 mm, had significant effects on the hardness, Ra, and Rz properties of the WPC materials.

• The highest hardness value of 79.65 was found in the milling process using an end mill diameter of 12 mm, whereas the minimum Ra and Rz values of 0.78 and 8.71 µm were found in WPC samples using an end mill of 8 mm.

• Increasing the speed parameter ranging from 220 to 720 rpm resulted in an evident increase in hardness values. On the other hand, a feed rate at 200 mm/min and a depth of cut at 3 mm slowly decreased the Ra and Rz values, resulting in necessary WPC surfaces with smooth appearances.

• This experimental result is different from the morphological structure and surface observation. It can be seen that the CNC milling condition using a high speed of 720 rpm displayed a smooth surface, which resulted in visible evenness on the WPC surfaces.

• The optimal condition of the CNC milling process using end mill diameters of 8 mm and 12 mm had a speed (A) of 720 rpm, a feed rate (B) of 220 mm/min, and a depth of cut (C) of 3 mm. The high-desirability score for overall properties was in agreement with the values of 0.973 (97.30%).

• This optimal condition of CNC milling parameters can also be used on other types of WPCs with different fibers and grades of PP or other plastics based on a melting point of more than 180°C.

The findings of the study also emphasized the fact that the manufacturing industry has advanced at a rapid pace. Consequently, high-performance manufacturing techniques become critical in attaining sustainable development.

Footnotes

Acknowledgements

The authors gratefully acknowledge Department of Industrial Engineering, Faculty of Engineering, Rajamangala University of Technology Srivijaya (Thailand) for the laboratory testing equipment for evaluating properties of the WPC materials. They would like to thank Mr Ponpawit Chairee and Mr Rungsiman Jantrasiri for supporting this work. Additionally, they would also like to thank Dr Somporn Maneechote and Dr Kittima Tantihachai for editing this article.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Chatree Homkhiew

Chainarong Srivabut

Data availability statement

The datasets used and/or analyzed during the current study available from the corresponding author on reasonable request.*

References

Khamtree

Homkhiew

Srivabut

, et al. Evaluation of mechanical properties of composites made from recycled plastic and waste rubberwood using multiply-response surface optimization as green building materials. Fibers Polym 2023; 24: 2819–2834. DOI: 10.1007/s12221-023-00266-w.

Hala

Kamal

Hamid

, et al. Mechanical and thermal properties of hybrid composites: oil-palm fiber/clay reinforced high density polyethylene. Mech Mater 2016; 98: 36–43. DOI: 10.1016/j.mechmat.2016.04.008.

Homkhiew

Srivabut

Boonchouytan

, et al. Optimization of the physical properties of recycled plastic hybrid composites reinforced with Rubberwood flour and crab shell flour for underwater use. Compos Commun: Open Access 2023; 12: 100411. DOI: 10.1016/j.jcomc.2023.100411.

Ratanawilai

Leelasilapasart

Srivabut

, et al. Lifetime prediction under dead-load long-term creep test and models analysis of wood-plastic composites for building materials. Plast, Rubber Compos 2023; 52: 506–515. DOI: 10.1080/14658011.2023.2232666.

Huang

Zhang

Chen

, et al. Thermal stability and flame resistance of the coextruded wood-plastic composites containing talc-filled plastic shells. International J Polym Sci 2020; 23: 1435249. DOI: 10.1155/2020/1435249.

Cheewawuttipong

Homkhiew

Rawangwong

. A comparative study on the effect of oil palm fiber contents and types on properties of rubberwood sawdust-polypropylene composites. Rajamangala Universi Technol Srivijaya Res J 2022; 14(1): 31–46, Retrieved from https://li01.tci-thaijo.org/index.php/rmutsvrj/article/view/242810

Husseinsyah

Chun

Hadi

, et al. Effect of filler loading and coconut oil coupling agent on properties of low-density polyethylene and palm kernel shell eco-composites. J Vinyl Addit Technol 2016; 22: 200–205. DOI: 10.1002/vnl.21423.

Srivabut

Homkhiew

Rawangwong

, et al. Possibility of using municipal solid waste for manufacturing wood-plastic composites: effects of natural weathering, wood waste types, and contents. J Mater Cycles Waste Manag 2022; 24: 1407–1422. DOI: 10.1007/s10163-022-01443-4.

Kaymakci

Birinci

Ayrilmis

. Surface characteristics of wood polypropylene nanocomposites reinforced with multi-walled carbon nanotubes. Compos B Eng 2019; 157: 43–46. DOI: 10.1016/j.compositesb.2018.08.099.

10.

O-Charoen

Srisawat

Jansri

. Effect of the particle size of polypropylene with flakes shape on rotational molding. Rajamangala Universi Technol Srivijaya Res J 2023; 15(3): 644–658, Retrieved from https://li01.tci-thaijo.org/index.php/rmutsvrj/article/view/252182

11.

Homkhiew

Srivabut

Ratanawilai

, et al. Characterization of polypropylene composites using sludge waste from the natural rubber latex industry as reinforcing filler. J Mater Cycles Waste Manag 2023; 25: 1444–1456. DOI: 10.1007/s10163-023-01621-y.

12.

Boussetta

Charii

Benhamou

, et al. Bio-composites based on cellulosic fbers from agro-industrial waste flled PP matrix: production and properties. Polym Bull 2023; 80: 13025–13050. DOI: 10.1007/s00289-023-04698-5.

13.

Chan

Pratt

Halley

, et al. Mechanical and physical stability of polyhydroxyalkanoate (PHA)-based wood plastic composites (WPCs) under natural weathering. Polym Test 2019; 73: 214–221. DOI: 10.1016/j.polymertesting.2018.11.028.

14.

Kruzelak

Kvasnicakova1

Hlozekova1

, et al. Experimental investigation of absorption shielding efficiency of rubber composites. Polym Bull 2023; 80: 13051–13068. DOI: 10.1007/s00289-023-04684-x.

15.

Liu

Wang

Gong

, et al. Morphology and mechanical properties of PVC/straw-fiber coated with liquid nitrile-butadiene rubber composites. J Appl Polym Sci 2016; 133: 44119. DOI: 10.1002/app.44119.

16.

Phiri

Mpitso

, et al. Curing, thermal and mechanical properties of waste type derived reclaimed rubber-wood flour composites. Mater Today Commun 2020; 25: 101204. DOI: 10.1016/j.mtcomm.2020.101204.

17.

Srivabut

Khamtree

Homkhiew

, et al. Comparative effects of different coastal weathering on the thermal, physical, and mechanical properties of rubberwood–latex sludge flour reinforced with polypropylene hybrid composites. Compos Commun: Open Access 2023; 12: 100383.

18.

Zhu

Buck

Wang

, et al. Machinability of different wood-plastic composites during peripheral milling. Materials 2022; 15: 1303. DOI: 10.3390/ma15041303.

19.

Shang

Ding

, et al. Particularities and research progress of the cutting machinability of wood-plastic composites. Mater Today Commun 2023; 37: 106924. DOI: 10.1016/j.mtcomm.2023.106924.

20.

Zhu

Buck

Guo

, et al. Energy Efficiency optimization for machining of wood plastic composite. Machines 2022; 10: 104. DOI: 10.3390/machines10020104.

21.

Bharathi

SRS

Ravindran

Moshi

AAM

, et al. Multi objective optimization of CNC turning process parameters with Acrylonitrile Butadiene Styrene material. Mater Today Proc 2020; 27: 2042–2047. DOI: 10.1016/j.matpr.2019.09.055.

22.

Mitalova

Botko

Vandzura

, et al. Machining of wood plastic composite using AWJ technology with controlled output quality. Machines 2022; 10: 566. DOI: 10.3390/machines10070566.

23.

Madic

Markovic

Radovanovic

. Optimization of surface roughness when turning polyamide using ANN-IHSA approach. Int J Eng Technol 2012; 1: 432–443. DOI: 10.14419/ijet.v1i4.378.

24.

Yang

Wang

. Modeling and predicting the machined surface roughness and milling power in Scot’s pine helical milling process. Machines 2022; 10: 331. DOI: 10.3390/machines10050331.

25.

Rabiei

Yaghoubi

. A comprehensive investigation on the influences of optimal CNC wood machining variables on surface quality and process time using GMDH neural network and bees optimization algorithm. Mater Today Commun 2023; 36: 106482. DOI: 10.1016/j.mtcomm.2023.106482.

26.

Guo

Ekevad

Marklund

, et al. Cutting forces and chip morphology during wood plastic composites orthogonal cutting. Bio 2014; 9: 2090–2106.

27.

Pei

Zhu

Gong

. A study on cutting temperature for wood–plastic composite. J Thermoplast Compos Mater 2009; 29: 1627–1640. DOI: 10.1177/089270571557098.

28.

Ibrahim

Hamni

Burhanudin

, et al. Optimization of cutting parameters on tool wear using Box-Behnken design. Int J Eng Sci 2019; 6: 15–23. DOI: 10.15282/ijets.6.1.2019.1002.

29.

Ray

Haque

Ahmed

, et al. Comparison of artificial neural network (ANN) and response surface methodology (RSM) in predicting the compressive and splitting tensile strength of concrete prepared with glass waste and tin (Sn) can fiber. J King Saud Univ Eng Sci 2023; 35: 185–199. DOI: 10.1016/j.jksues.2021.03.006.

30.

Saeb

Moghri

. Application of response surface methodology in describing fusion characteristics of rigid poly(vinyl chloride)/nanoclay/wood flour/calcium carbonate hybrid nanocomposites. J Vinyl Addit Technol 2014; 34: 183–190. DOI: 10.1002/vnl.21387.

31.

Srivabut

Ratanawilai

Hiziroglu

. Response surface optimization and statistical analysis of composites made from calcium carbonate filler-added recycled polypropylene and rubberwood fiber. J Thermoplast Compos Mater 2022; 35: 391–415. DOI: 10.1177/0892705719889988.

32.

Khamtree

Ratanawilai

. The effect of alkaline-silane treatment of rubberwood flour for water absorption and mechanical properties of plastic composites. J Thermoplast Compos Mater 2018; 33: 599–613. DOI: 10.1177/089270571880855.

33.

Ibitoye

Radford

. Experimental and statistical study on the effect of process parameters on the quality of continuous fiber composites made via additive manufacturing. J Thermoplast Compos Mater. 2024. DOI: 10.1177/089270572412415.

34.

Patel

Chaudhary

Patel

, et al. Milling of polymer matrix composites: a review. Int J Appl Eng 2018; 13: 7455–7465.

35.

Slamani

Chafai

Chatelain

. Effect of milling parameters on the surface quality of a flax fiber-reinforced polymer composite. Proc Inst Mech Eng 2022; 13: 1–8. DOI: 10.1177/09544089221126087.

36.

Yang

Guo

Wan

, et al. Surface roughness prediction and optimization in the orthogonal cutting of graphite/polymer composites based on artificial neural network. Processes 2021; 9: 1858. DOI: 10.3390/pr9101858.

37.

Doluk

Rudawska

Miturska-Baranska

. Investigation of the surface roughness and surface uniformity of a hybrid sandwich structure after machining. Materials 2021; 15: 7299. DOI: 10.3390/ma15207299.

38.

Lmalghan

Rao

KMC

Arun

, et al. Machining parameters optimization of AA6061 using response surface methodology and particle swarm optimization. Int J Precis Eng Manuf 2018; 19: 695–704. DOI: 10.1007/s12541-018-0083-2.

39.

Geier

Pereszlai

. Analysis of characteristics of surface roughness of machined CFRP composites. Period Polytech - Mech Eng 2020; 64: 67–80. DOI: 10.3311/PPme.14436.

40.

Wei

Shang

Peng

, et al. Prediction model of sound signal in high-speed milling of wood-plastic composites. Materials 2022; 15: 3838. DOI: 10.3390/ma15113838.

41.

Hrbackova

Sedlak

Chromjakova

, et al. Evaluation of the effect of machining technologies on the surface texture analysis of ertacetal C polymer. Manuf Technol 2022; 22: 679–692. DOI: 10.21062/mft.2022.083.