Multi-objective optimization of WC-12Co coating by high-velocity oxygen fuel spray using multiple regression-based weighted signal-to-noise ratio

Abstract

In this study, the optimal process parameters for a high-velocity oxygen fuel spray, which was used to deposit WC-12Co coating on a 16Mn steel substrate, were studied to simultaneously improve the microhardness, adhesion strength, and porosity of the coating. To this end, trial tests were conducted to determine the feasible ranges of the spray parameters: powder feed rate (A), spray distance (B), and oxygen/propane mixture ratio (C). Taguchi orthogonal array (L9) experimental design was used to determine regression formulae for the microhardness, adhesion strength, and porosity. Then, based on a method using the multiple regression-based weighted signal-to-noise ratio, the optimal spray parameters were identified. To verify the optimization results, confirmation experiments were carried out. An analysis of variance was also conducted to check the proportion of contribution of the spray parameters on the output quality characteristics. The confirmation test results proved that the predicted and measured results were in good agreement. The optimal conditions of the high-velocity oxygen fuel spray were: A = 32 g/min, B = 0.35 m, and C = 5. The estimated optimal coating properties were: microhardness = 1335.784 HV, adhesion strength = 64.659 MPa, and porosity = 1.797%. The oxygen/propane ratio showed the highest contribution to the microhardness, whereas the most influential factor for both adhesion strength and porosity was the spraying distance.

Keywords

Adhesion strength HVOF spray microhardness multiple regression-based weighted signal-to-noise ratio porosity WC-12Co

Introduction

Thermal spray techniques have been used extensively to improve the surface properties and dimensional restoration of metal parts by means of deposited coatings. Of these techniques, high-velocity oxygen fuel (HVOF) spraying is highly flexible and suitable for a wide range of substrate materials, such as alloys, metallic ceramics, and carbides.¹ Low alloy, high-strength 16Mn steel has been extensively used in many industries, especially in mining machinery, and civil construction. Many mechanical components made of 16Mn steel experience severe friction and wear during their operation, such as drilling rigs, the bucket teeth of hydraulic excavators, crusher shafts, etc. Hence, deposited erosion-resistant coatings made with HVOF spray for those components are extremely important to extend their service life and restore their dimensions.

The HVOF coatings have the advantages of low porosity, and high adhesion strength and hardness compared with other spraying methods.² However, the performance of the HVOF coatings is highly sensitive to the spray process parameters.³ Therefore, the determination of optimal spray parameter values is critical for obtaining superior coating properties, and has attracted significant attention from researchers. Among the optimization techniques, the design of experiments (DOE) using the Taguchi approach is fairly effective for finding the optimal process parameters for many industrial implementations, including the thermal spray process.^4–8 Praveen et al.⁴ used the Taguchi method for the optimization of HVOF spray with NiCrSiB/WC-Co coating on an AISI 304 stainless steel substrate. Qiao et al.⁵ studied the optimal parameters of HVOF spray using an iron-based amorphous coating on a 45-steel substrate also using the Taguchi method. Wang et al.⁶ examined the abrasive wear response of a WC-Co coating deposited on a 16Mn steel substrate, in which kerosene was used in the HVOF spray. Qin et al.⁷ explored the optimal spraying parameters of HVOF spray with Fe-based coatings. The effects of the main process parameters on residual stresses in HVOF sprayed WC-12Co coatings were studied by Nouzouri et al.⁸

The Taguchi approach is normally suitable for the optimization of a single objective. To solve the multi-objective optimization of thermal spray processes, several authors have developed the Taguchi-OEC technique,⁹ Taguchi-based gray relational analysis^10,11 and response surface methodology.^12,13 Pal and Gauri¹⁴ introduced the multiple regression-based weighted signal-to-noise ratio (MRWSN) optimization method, which was based on a combination of the Taguchi and mathematical regression techniques. They applied this method to various optimization problems in engineering such as plasma-enhanced chemical vapor deposition, wire electrical discharge machining, and polysilicon deposition.¹⁵ The noticeable advantages of this method are its easy implementation and low computational complexity. Nevertheless, the MRWSN method still can offers competitive computational accuracy in comparison with other methods.¹⁶ Therefore, it may be useful for process engineers.

In this work, the multi-objective optimization of the HVOF spray process with a WC-12Co coating and 16Mn steel substrate was studied. The objectives of optimization were the microhardness, adhesion strength, and porosity of the coating. The spray parameters to be optimized were the powder feed rate, spraying distance, and oxygen/propane ratio. Trial tests were prepared to determine the feasible ranges of the spray parameters. Then, each spray parameter range was divided into three levels, and spray experiments were conducted. The optimal spray parameters and correspondent quality characteristics-based multi-objective optimization were determined using the MRWSN method. In order to check the optimal results determined by theoretical computation, confirmation experiments were conducted. Finally, an analysis of variance (ANOVA) was carried out to determine the degree of influence of each spray parameter on the output measures, and the obtained results are discussed.

Materials and experiment method

Materials

The substrate material was 16Mn steel with the chemical composition shown in Table 1. The powder used in the HVOF spray was WC-12Co, provided by Eutectic Corp. (USA). Figure 1 shows the morphology of the WC-12Co powder, which was captured by scanning electron microscopy (SEM). It can be seen from Figure 1 that the WC-12Co powder is spherical in shape with particle sizes ranging from 15 to 45 µm. Moreover, powder particles with a porous structure can be observed. Figure 2 presents the energy-dispersive X-ray (EDX) scanning pattern of the WC-12Co powder. From Figure 2, the chemical composition of the powder was identified as depicted in Table 2.

Table 1.

Chemical composition of the substrate material, 16 Mn steel.

Element	C	Si	Mn	P	S	V	Nb	Ti
wt. %	0.12–0.18	0.2–0.55	1.2–1.6	≤0.04	≤0.04	0.02–0.15	0.015–0.06	0.02–0.2

Figure 1.

Morphology of the WC-12Co powder: (a) SEM image of WC-12Co powder and (b) enlarged view of a particle.

Figure 2.

EDX scanning results of the WC-12Co powder.

Table 2.

Chemical composition of the WC-12Co powder.

Element	Co	C	S	O	W
wt. %	13.27	5.45	0.07	1.62	Remaining

Experiment method

The dimensions of the specimens used in the HVOF spray were 50×50×6 mm. The specimens were cut from an original steel plate by using an electrical discharge machine, which was intended to minimize the effect of the cutting process on the steel substrate properties. Subsequently, the specimens were blasted by an abrasive blasting machine (TSA–206, Vietnam) to increase the adhesion strength between the deposition and the substrate. The size of the sand particles used in the blasting was between 0.8 and 1 mm. The surface roughness of the specimens after blasting varied between 8 and 10 μm, and was checked using a surface roughness gage (SurTest TR100, ElektroPhysik, Germany). After blasting, the specimens were cleaned with Acetone solution to remove the embedded particles, and were subsequently air dried at 45°C for 5 min. The WC-12Co powder was dried in a vacuum oven for 2 h at 120 to 150°C to remove the absorbed moisture.

The spraying process was performed on a Hipo-Jet HVOF system, which consisted of a MP-2100 control panel, an HP-2700M spray gun, and a PF-3350 powder feeder. Before spraying, the specimen was pre-heated using the spray gun at 150°C for 2 min. The pre-heating temperature was checked by a non-contact thermometer. After that, the spray process commenced. The required thickness of the deposited coating after the spraying process was 500 ± 50 µm. Figure 3 presents the specimens before and after spraying.

Figure 3.

The specimens before spraying (on the left) and after spraying (on the right).

Coating properties characterization

This section presents the characterization methods used to measure the microhardness, adhesion strength, and porosity of the coating. The microhardness was measured using a hardness tester (IndentaMet 1106, Buhler, USA) based on the ASTM E384-17:2011 standard.¹⁷ During the microhardness test, a load of 100 g was applied using an indenter, and the elapsed time of indentation was 15 s. Then, microhardness was measured on the HV_0.1 scale. Figure 4 shows the specimen after the microhardness test.

Figure 4.

SEM image of specimen after the microhardness test.

The adhesion strength of the coating was characterized according to the JIS-H-8666-1980 standard.¹⁸ The dimensions of the specimens used in the adhesion strength test were 7×50×6 mm. The specimens were cut from the initial sample with the size of 50×50×6 mm by using an electrical discharge machine. Figure 5 shows the experimental setup for measuring the adhesion strength. A custom-designed fixture was fabricated to perform the test. Figure 5 also illustrates the schematic diagram of the designed fixture. The fixture was fixed to the test bed of a universal testing machine, which generated a compressive load on the punch. The load was monitored, and the adhesion strength of the coating was calculated as follows:

Adhesion strength = \frac{F}{S}

(1)

where F is the measured load in newtons (N). S is the shearing area, which was equal to the embedded area of the coating on the substrate: S = 50×7 mm².

Figure 5.

Experiment setup for measuring the adhesion strength.

The porosity test was guided by the ASTM B276-05-2015 standard.¹⁹ Accordingly, the test was conducted on a 400×400 µm square area of a cross section over the coating thickness. The porosity was theoretically calculated as the ratio of the total areas of the voids to the examined square area. Images of well-polished cross-sections were captured using the inverted microscope Axioplan 2 (Carl Zeiss, Germany) at a 200× magnification. The solid background of the magnified image was removed to determine the porosity, which was done using the image processing program ImageJ (NIH, USA). Figure 6 demonstrates the magnified images from the porosity test.

Figure 6.

Porosity test with the original image and the image with solid background removed.

The measurement of the coating characteristics was repeated at least five times to ensure reproducibility. The final measured value was the average of the recorded measurements.

Experimental design

With HVOF spray, the powder feed rate, spray distance, and oxygen/propane ratio are the critical parameters that predominantly affect the quality of the deposited coating.^9,20,21

To identify the feasible range of the above-mentioned spray parameters, a number of trial tests were carried out with varying powder feed rates, spray distances, and oxygen/propane ratios. The other parameters, which were kept constant during the process, are as follows: oxygen and propane pressure = 0.686 MPa, propane flow rate = 60 L/min, carrier gas (nitrogen) pressure = 0.4 MPa, carrier gas flow rate = 20 L/min. From the testing trials, some observations of the deposited microstructures were obtained, as summarized in Table 3. From Table 3, the working ranges of the spray parameters were selected as follows: powder feed rate from 26 to 38 (g/min), spray distance from 0.2 to 0.35 (m), and oxygen/propane ratio from 4 to 6. To find the regression formulae for the response qualities, a total of nine experiments were conducted. The experiments followed the L9 orthogonal array recommended by Taguchi. Three levels were selected for each spray parameter, including powder feed rate (A), with A₁ = 26, A₂ = 32, and A₃ = 38 (g/min); spray distance (B), with B₁ = 0.2, B₂ = 0.275, and B₃ = 0.35 (m); and oxygen/propane ratio (C), with C₁ = 4, C₂ = 5, and C₃ = 6. Table 4 shows the details of the experiment plan.

Table 3.

Testing trial observations.

Microstructure	Observations
Powder feed rate < 26 g/min
	Ineffective deposition for spraying at high spray distance, defective microstructure with voids.
Powder feed rate > 38 g/min
	Unmelted powder particles, high porosity, defectivemicrostructure with voids.
Spray distance < 0.2 m
	Overheating of the substrate, high porosity coating, defects such as gap at boundaries between substrate and coating.
Spray distance > 0.35 m
	High porosity.
Oxygen/propane ratio < 4
	Unmelted powder particles, high porosity.
Oxygen/propane ratio > 6
	High porosity.

Table 4.

Experiment plan for HVOF spray.

Exp. number	Powder feed rate (A)		Spraying distance (B)		Oxygen/propane ratio (C)
	Value (g/min)	Level	Value (m)	Level	Value	Level
1	26	1	0.200	1	4	1
2	26	1	0.275	2	5	2
3	26	1	0.350	3	6	3
4	32	2	0.200	1	5	2
5	32	2	0.275	2	6	3
6	32	2	0.350	3	4	1
7	38	3	0.200	1	6	3
8	38	3	0.275	2	4	1
9	38	3	0.350	3	5	2

Multi-objective optimization methodology

MRWSN method

This section presents the calculation process for the MRWSN method for the multi-objective optimization. First, the average values of response quality $(y)$ and covariance $(σ^{2})$ for all the experiments were calculated using the following equations:

y = \frac{\sum_{k = 1}^{n} y_{k}}{n}

(2)

σ^{2} = \sum_{k = 1}^{n} {(y_{k} - y)}^{2}

(3)

where y indicates the value of response quality (microhardness, adhesion strength, and porosity), and n is the number of repeated measurements for one experiment.

Next, regression formulae for response quality $y$ and $lo g_{10} (σ^{2})$ as a function of spray parameters were developed using a least-squares method.²² A detailed determination of the regression formulae is presented in section 4.2. After the $y$ and $lo g_{10} (σ^{2})$ formulae were obtained, the S/N ratios were determined. For the smaller-is-better criterion, which was used for coating porosity, the S/N ratio was calculated by:

S / N = - 10 \log_{10} (y^{2} + σ^{2})

(4)

For the larger-is-better criterion, which was used for the microhardness and adhesion strength of the coating to the substrate, the S/N ratio was determined as:

S / N = - 10 \log_{10} [\frac{1}{y^{2}} (1 + 3 \frac{σ^{2}}{\bar{y^{2}}})]

(5)

From the calculated S/N ratios, the MRWSN was estimated using the following equation:

MRWSN = \sum_{j = 1}^{p} w_{j} \times S / N_{j}

(6)

where S/N_j represents the signal to noise ratio for the j-th response quality. Here, w_j is the weighting factor of the response quality, which satisfies $\sum_{j = 1}^{p} w_{j} = 1$ . In this study, for the three response qualities of microhardness, adhesion strength, and porosity (p = 3), the weighting factors were respectively selected as follows: w₁ = 0.5, w₂ = 0.4, and w₃ = 0.1.¹³ It should be noted that there were a total of 27 values for MRWSN, which corresponded to 27 sets of spray parameters based on the MRWSN method.¹⁴

After attaining all the MRWSN values, the set of spray parameters corresponding to the highest value of MRWSN were identified. Those parameters represented the optimal spray conditions. Finally, the response qualities associated with the optimal parameters were calculated using the regression equations found in the previous step.

Determination of regression formula

The general form of the regression formula is:

\begin{matrix} L = a_{0} + a_{1} A + a_{2} B + a_{3} C + a_{4} AB + a_{5} AC \\ + a_{6} BC + a_{7} A^{2} + a_{8} B^{2} + a_{9} C^{2} \end{matrix}

(7)

where A, B, and C are the spray parameters. a_j (j = 1–9) are the coefficients to be determined. These coefficients are determined by minimizing the following residual sum of squares SSR:

SSR = \sum_{i = 1}^{N} {(L_{i} - y_{i})}^{2} \to minimized

(8)

where subscript i indicates the experiment. N is total number of experiments, N = 10. Accordingly, the coefficients a_j (j = 1–9) are the roots of following system of equations:

{\begin{matrix} \frac{\partial SSR}{\partial a_{0}} = 0 \\ \frac{\partial SSR}{\partial a_{1}} = 0 \\ \dots \dots \dots \\ \frac{\partial SSR}{\partial a_{9}} = 0 \end{matrix}

(9)

The above system of equations was transformed to the following standard form as:

{\begin{matrix} \sum_{i = 1}^{N} L_{i} = a_{0} \sum_{i = 1}^{N} 1 + a_{1} \sum_{i = 1}^{N} A_{i} \\ + a_{2} \sum_{i = 1}^{N} B_{i} + \dots + a_{9} \sum_{i = 1}^{N} C_{i}^{2} \\ \sum_{i = 1}^{N} A_{i} L_{i} = a_{0} \sum_{i = 1}^{N} A_{i} + a_{1} \sum_{i = 1}^{N} A_{i}^{2} \\ + a_{2} \sum_{i = 1}^{N} A_{i} B_{i} + \dots + a_{9} \sum_{i = 1}^{N} A_{i} C_{i}^{2} \\ \dots \dots \dots \dots \dots \dots \dots \dots \\ \sum_{i = 1}^{N} C_{i}^{2} L_{i} = a_{0} \sum_{i = 1}^{N} C_{i}^{2} + a_{1} \sum_{i = 1}^{N} C_{i}^{2} A_{i} \\ + a_{2} \sum_{i = 1}^{N} C_{i}^{2} B_{i} + \dots + a_{9} \sum_{i = 1}^{N} C_{i}^{4} \end{matrix}

(10)

By solving the system of equations in (10), the coefficients a_j (j = 1–9) were found. To solve this system of equations, an iterative Newton–Raphson method was used. After that, a goodness of fit test was performed to confirm the accuracy of the derived regression equations. To perform the goodness of fit test, the sum of squares explained by the regression (SSR) was calculated using equation (8). Subsequently, the total sum of squares SST, which represents the total difference between each variable and its mean value, was calculated:

SST = \sum_{i = 1}^{M} {(y_{i} - \bar{y_{i}})}^{2}

(11)

where $y_{i}$ is the measured value in the i-th experiment. $\bar{y_{i}}$ indicates the mean value of the i-th experiment calculated by the regression equation. Finally, the coefficient of determination ( $R^{2}$ ) was computed:

R^{2} = 1 - \frac{SSR}{SST}

(12)

An $R^{2}$ very close to 1.0 will ensure the goodness of the predicted formulae with small error.

Results and discussion

Experimental results

Table 5 shows the experimental results for microhardness, adhesion strength, and porosity. Table 5 demonstrates that the microhardness, adhesion strength, and porosity of the coatings vary from 1058.2 to 1255.5 HV, 52.1 to 67.6 MPa, and 1.42 to 3.37%, respectively.

Table 5.

Experiment results for coating properties obtained by HVOF spray.

Exp.number	Microhardness(HV)	Adhesionstrength (MPa)	Porosity (%)
1	1058.2 ± 69.02	52.1 ± 1.90	3.18 ± 0.20
2	1166.8 ± 175.97	65.3 ± 0.71	2.16 ± 0.15
3	1167.3 ± 128.03	61.2 ± 1.41	1.42 ± 0.22
4	1255.5 ± 119.95	59.5 ± 0.82	2.79 ± 0.31
5	1180.5 ± 115.66	65.6 ± 0.28	2.14 ± 0.19
6	1229.8 ± 124.47	58.3 ± 1.44	2.30 ± 0.21
7	1095.6 ± 128.14	53.8 ± 0.37	2.96 ± 0.19
8	1018.9 ± 163.41	55.0 ± 1.19	3.37 ± 0.27
9	1245.4 ± 157.07	57.5 ± 1.27	1.92 ± 0.18

Figure 7 shows the microstructure of the coatings with the largest (Experiment 1) and smallest porosity (Experiment 3). Experiment 3 has a higher spraying distance and a higher oxygen/propane ratio than Experiment 1, while the powder feed rates were the same. The deposited coatings generally show dense microstructures. However, there are voids in the cross-section of the coating for Experiment 1 as shown in Figure 7(a). A coating with denser microstructure is generally formed with a lower porosity (lesser volume of voids). In HVOF spray, the energy absorbed by the powder particles has a great impact on the quality of coating. This is because only the particles with sufficient energy will experience a plastic deformation and fully transform to a molten state. The energy gained by the particles also influences the solidification behavior of the molten particles. The small powder particles underwent a full phase transformation from a solid to a melt state in Experiment 1. However, a partially melted state occurred in the larger particles because the particles gained insufficient energy during the short spraying time. As a result, voids formed in the deposited coating, as can be observed in Figure 7(a). On the other hand, the deposited matrix shows complete melting when the spraying distance is 0.35 m (Experiment 3). Therefore, the particles underwent a full transition from solid to melt phase, which led to a reduction in coating defects, as shown in Figure 7(b).

Figure 7.

Microstructure of coatings with minimal and maximal porosity: (a) experiment 1 and (b) experiment 3

Figure 8 shows the interface regions between the coating and substrate for Experiments 1 and 3. The characteristics of the interface have a major impact on the strength of the adhesion between the substrate and coating. Defects such as pores and gaps are found at the boundaries, as shown for Experiment 1 in Figure 8(a). With an increasing oxygen/propane ratio, the thermal source produced by the spray gun increases. Hence, the coating powder can be deposited with a higher degree of melting to form good bonding with the substrate, which contributes to an increase in the adhesion strength.

Figure 8.

SEM image of the interface between coating and substrate: (a) experiment 1 and (b) experiment 3.

The above-mentioned improvement in the melting ability of the particles also accounts for the increased microhardness in Experiment 3 compared with Experiment 1, as shown in Table 5. Nevertheless, an improvement in coating qualities cannot be attained by an increase or decrease in individual spray parameters. Instead, the coating qualities are dependent on how the three spray parameters are combined. Hence, it is essential to find the optimal combination of spray parameters.

Figure 9 shows the typical phase compositions of the coating obtained using EDX line scanning. Compared with the phase compositions of the original powder, new phases were formed, such as W2C and Co4W2C, caused by the phase transformation process due to heat and also the quick cooling process of the HVOF spray. Table 6 lists the main chemical compositions of the deposition. The change in spray parameters resulted in a subsequent change in the chemical composition of the coating, as shown in Table 6.

Figure 9.

Phase composition of coating.

Table 6.

Main chemical compositions of the coating (wt. %).

Element	WC	W2C	W	Co4W2C
Experiment 3	29.95	13. 52	9.91	13.55
Experiment 8	35.29	6.76	6.65	9.09

Multi-objective optimization

This section presents the determination of the optimal spraying parameters using the MRWSN approach. From the experimental results represented in Table 5, the regression formulae for $y$ and $lo g_{10} (σ^{2})$ were determined using the least-squares method.²² The regression formulae for each response quality were found.

For the microhardness:

\begin{matrix} y = - 3088.39 + 147.99 A - 4394.61 B + 982.78 C \\ + 24.9 AB + 0.487 AC - \\ 185.42 BC - 2.51 A^{2} + 9205.33 B^{2} - 92.0 C^{2} \end{matrix}

(13)

\begin{matrix} \log_{10} (σ^{2}) = 11.54 - 1.636 A - 90.19 B + 12.75 C \\ + 4.312 AB - 0.357 AC - \\ 28.98 BC + 0.0325 A^{2} + 153.6 B^{2} + 0.765 C^{2} \end{matrix}

(14)

For the adhesion strength:

\begin{matrix} y = - 207 + 6.649 A + 631.926 B + 30.657 C \\ - 3.185 AB + 0.206 AC + 9.333 BC - \\ 0.11 A^{2} - 986.667 B^{2} - 3.8 C^{2} \end{matrix}

(15)

\begin{matrix} \log_{10} (σ^{2}) = 5.553 + 0.42 A - 9.584 B - 3.982 C \\ - 2.076 AB + 0.122 AC + 14.65 BC - \\ 0.0062 A^{2} + 19.04 B^{2} + 0.487 C^{2} \end{matrix}

(16)

For the porosity:

\begin{matrix} y = 8.559 + 0.067 A + 15.141 B - 3.196 C - 0.452 AB \\ + 0.015 AC + \\ 1.911 BC + 0.0006 A^{2} - 30.815 B^{2} + 0.171 C^{2} \end{matrix}

(17)

\begin{matrix} \log_{10} (σ^{2}) = - 0.296 + 0.137 A - 20.88 B + 0.024 C \\ + 0.269 AB - 0.038 AC - \\ 1.025 BC - 0.00024 A^{2} + 28.29 B^{2} + 0.15 C^{2} \end{matrix}

(18)

Table 7 shows the goodness of fit test results for the regression formulae (13)–(18). It can be observed that the calculated values of the coefficient of determination ( $R^{2}$ ) are all higher than 0.97. Therefore, the results calculated by the formulae are well fit to the measurements, with minimal error. Thus, the formulae could be further used for the optimization design.

Table 7.

Goodness of fit test results for the regression formulae for y and $\log_{10} (σ^{2})$ .

Formulae	y			$\log_{10} (σ^{2})$
	SSR	SST	R ²	SSR	SST	R ²
Microhardness	6084.067	1366451.365	0.996	0.097	26.645	0.996
Adhesion strength	6.117	3832.383	0.998	0.005	2.426	0.998
Porosity	0.016	9.144	0.998	0.025	0.852	0.971

Based on the MRWSN theory, a combination of 27 sets of spray parameters were created, as shown in column 2 of Table 8. From equations (13)–(18), the $y$ and $σ^{2}$ values for all the parameter sets were calculated. Afterward, the S/N ratios and MRWSN were determined using equations (4)–(6). Table 8 shows the computed values of the S/N ratios and MRWSN. From this table, the highest value for MRWSN was identified from A₂B₃C₂, that is, Experiment 17. The optimal spray parameters were found to be as follows: A=32 g/min, B=0.35 m, and C=5. The corresponding response characteristics were computed by substituting the optimal parameters into equations (13), (15), and (17). The optimal response characteristics were then obtained: microhardness=1335.784 HV, adhesion strength = 64.659 MPa, and porosity = 1.797%.

Table 8.

Calculated values of S/N ratios and MRWSN.

Exp. number	Spray parameters	S/N ratio for			MRWSN	Rank
		Microhardness	Adhesion strength	Porosity
1	A ₁ B ₁ C ₁	60.237	34.292	−10.105	42.825	20
2	A ₁ B ₁ C ₂	29.153	34.934	−7.350	27.815	24
3	A ₁ B ₁ C ₃	−32.536	34.300	−5.594	−3.108	27
4	A ₁ B ₂ C ₁	60.410	35.628	−9.306	43.525	16
5	A ₁ B ₂ C ₂	60.144	36.299	−6.757	43.916	12
6	A ₁ B ₂ C ₃	26.143	35.863	−5.128	26.904	25
7	A ₁ B ₃ C ₁	61.160	35.111	−7.253	43.899	13
8	A ₁ B ₃ C ₂	61.909	35.898	−4.629	44.851	3
9	A ₁ B ₃ C ₃	60.675	35.721	−3.314	44.295	6
10	A ₂ B ₁ C ₁	60.989	34.469	−11.208	43.161	18
11	A ₂ B ₁ C ₂	61.458	35.492	−9.049	44.021	8
12	A ₂ B ₁ C ₃	30.769	35.096	−7.618	28.661	23
13	A ₂ B ₂ C ₁	61.005	35.803	−9.973	43.826	14
14	A ₂ B ₂ C ₂	61.885	36.590	−7.928	44.786	4
15	A ₂ B ₂ C ₃	60.902	36.346	−6.781	44.311	5
16	A ₂ B ₃ C ₁	61.217	35.298	−7.403	43.987	10
17	A ₂ B ₃ C ₂	62.509	36.125	–5.211	45.183	1
18	A ₂ B ₃ C ₃	61.993	35.991	−4.372	44.956	2
19	A ₃ B ₁ C ₁	60.032	33.093	−12.278	42.025	22
20	A ₃ B ₁ C ₂	61.140	34.760	−10.592	43.415	17
21	A ₃ B ₁ C ₃	60.047	34.627	−9.586	42.916	19
22	A ₃ B ₂ C ₁	58.838	34.817	−10.712	42.275	21
23	A ₃ B ₂ C ₂	61.134	35.822	−9.090	43.987	11
24	A ₃ B ₂ C ₃	60.668	35.736	−8.355	43.793	15
25	A ₃ B ₃ C ₁	26.238	33.986	−7.781	25.935	26
26	A ₃ B ₃ C ₂	61.048	35.187	−5.948	44.004	9
27	A ₃ B ₃ C ₃	61.363	35.147	−5.525	44.188	7

Experimental verification

In order to verify the computed optimization results, confirmation experiments were conducted with the optimal spray parameters, A = 32 g/min, B = 0.35 m, and C = 5, that were determined using the theoretical computation. Then, the response characteristics were measured and compared to those obtained by prediction.

Table 9 shows the verification results. This table clearly shows that the measured and predicted values of microhardness, adhesion strength, and porosity are in a good agreement. The errors between measurement and prediction are smaller than 4% for all the response characteristics. Hence, the accuracy of the optimal results found using the MRWSN method was confirmed.

Table 9.

Confirmation test result.

Output measure	Prediction	Experiment	Error (%)
Microhardness (HV)	1335.784	1329.6	0.465
Adhesion strength (MPa)	64.659	65.3	0.982
Porosity (%)	1.797	1.73	3.873

ANOVA results

In this section, the ANOVA results are presented. The main objective of the ANOVA is to determine the degree of importance of a spray parameter (or factor) on each quality characteristic via a so-called “contribution percentage.” Accordingly, the most influential factor, which shows the maximum contribution percentage, can be identified.^23,24

Table 10 shows the results of the ANOVA. From this table, the ratio of oxygen/propane (C) exhibits the greatest influence on the microhardness, with a contribution of 39.4%. The next greatest influences are the powder feed rate (A) and spray distance (B), with contributions of 33.5 and 26.2%, respectively. On the other hand, the most influential factor that affects the adhesion strength and porosity is the spraying distance, with a contribution of 38.9 and 55.7% respectively. The powder feed rate shows the least effect on both adhesion strength and porosity, with contribution percentages of 28.7 and 11.7%, respectively.

Table 10.

The results of the ANOVA for the quality characteristics of HVOF spray.

Factor	S_l			Contribution (%)
	Microhardness	Adhesion strength	Porosity	Microhardness	Adhesion strength	Porosity
A	18825.2	51.860	0.38682	33.5	28.7	11.7
B	14731.3	70.447	1.83696	26.2	38.9	55.7
C	22124.4	57.727	1.05242	39.4	31.9	31.9
Error	490.4	0.887	0.01929	0.9	0.5	0.7
Total	56171.3	180.920	3.29549	100	100	100

Conclusions

In this study, experiments using HVOF spray with WC-12Co powder on a 16Mn steel substrate were conducted. The microhardness, adhesion strength, and porosity of the coating were characterized. The MRWSN method was adopted to simultaneously optimize the quality characteristics. The conclusions of this paper were as follows:

The feasible working ranges of the spray parameters confirmed by the trial test were: a powder feed rate from 26 to 38 (g/min), a spray distance from 0.2 to 0.35 (m), and an oxygen/propane ratio from 4 to 6.

The regression formulae for the three coating characteristics were found. Good results for the goodness of fit tests for the formulae were confirmed with all coefficients of determination (R²) > 0.97.

The optimal conditions of the HVOF spray were identified: powder feed rate A = 32 g/min, spray distance B = 0.35 m, and oxygen/propane ratio C = 5. The output measures predicted for the optimal spray parameters were microhardness = 1335.784 HV, adhesion strength = 64.659 MPa, and porosity = 1.797%, which agreed well with the measured results of the confirmation experiments, with errors less than 4%.

The ANOVA results showed the individual contributions of the spray parameters to the output coating characteristics. The oxygen/propane ratio showed the highest contribution to the microhardness, whereas the most influential factor for both adhesion strength and porosity was the spraying distance.

Footnotes

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) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Ministry of Education and Training, Vietnam (Grant No. B2021-SKH-02)

ORCID iD

Van-Canh Tong

References

Pawlowski

The science and engineering of thermal spray coatings. 2nd ed. West Sussex: John Wiley & Sons, 2008.

Tejero-Martin

Rezvani Rad

McDonald

, et al. Beyond traditional coatings: a review on thermal-sprayed functional and smart coatings. J Therm Spray Technol 2019; 28: 598–644.

Abu-warda

López

, et al. Ni20Cr coating on T24 steel pipes by HVOF thermal spray for high temperature protection. Surf Coat Technol 2020; 381: 125133.

Praveen

Sarangan

Suresh

, et al. Optimization and erosion wear response of NiCrSiB/WC–Co HVOF coating using Taguchi method. Ceram Int 2016; 42: 1094–1104.

Qiao

Hong

, et al. Wei, Influence of the high-velocity oxygen-fuel spray parameters on the porosity and corrosion resistance of iron-based amorphous coatings. Surf Coat Technol 2019; 366: 296–302.

Wang

Tang

, et al. Abrasive wear behavior of WC-Co coating deposited by high velocity oxygen flame process. Chin J Nonferr Metals 2015; 25: 1920–1928 (in Chinese).

Qin

Zhang

, et al. Optimization of the HOVF spray parameters by Taguchi method for high corrosion-resistant Fe-Based coatings. J Mater Eng Perform 2015; 24: 2637–2644.

Nourouzi

Azizpour

Salimijazi

HR.

Parametric study of residual stresses in HVOF thermally sprayed WC–12Co coatings. Mater Manuf Process 2014; 29: 1117–1125.

Dinh

Nguyen

Tong

VC.

Multi-response optimization of 67Ni18Cr5Si4B coating by HVOF spray using Taguchi-OEC technique. J Adhesion Sci Technol 2019; 33: 314–327.

10.

Thirumalvalavan

Senthilkumar

Optimising the wear performance of HVOF thermal spray coated Ti-6Al-4V alloy by grey relational approach. Int J Rapid Manuf 2020; 9: 25–47.

11.

Kumar

Pandey

NK.

Optimization of the process parameters in generic thermal barrier coatings using the Taguchi method and Grey relational analysis. Proc IMechE Part L: J Materials: Design and Applications 2017; 231: 600–610.

12.

Murugan

Ragupathy

Balasubramanian

, et al. Optimizing HVOF spray process parameters to attain minimum porosity and maximum hardness in WC-10Co-4Cr coatings. Surf Coat Technol 2014; 247: 90–102.

13.

Pierlot

Pawlowski

Bigan

Design of experiments in thermal spraying: a review. Surf Coat Technol 2008; 202: 4483-4490.

14.

Pal

Gauri

SK.

Multi-response optimization using multiple regression-based weighted signal-to-noise ratio (MRWSN). Qual Eng 2010; 22: 336–350.

15.

Pal

Gauri

SK.

Assessing effectiveness of the various performance metrics for multi-response optimization using multiple regression. Comput Ind Eng 2010; 59: 976–985.

16.

Peng

Yeh

Lai

, et al. The incorporation of the Taguchi and the VIKOR methods to optimize multi-response problems in intuitionistic fuzzy environments. J Chin Inst Eng 2015; 38: 897–907.

17.

ASTM International. ASTM-E384-17: standard test method for microindentation hardness of materials. USA: ASTM International, 2011.

18.

Japanese Industrial Standards. JIS-H-8664: test methods for build-up thermal spraying. Japan: Japanese Industrial Standards, 2004.

19.

ASTM International. ASTM-B276-05: standard test method for apparent porosity in cemented carbides. USA: ASTM International, 2015.

20.

Zouari

Ghorbel

Danlos

, et al. Comparative study of HVOF-sprayed NiCrBSi alloy and 316L stainless steel coatings on a brass substrate. J Therm Spray Technol 2019; 28: 1284–1294.

21.

Jiaxing

Jie

Dejun

Microstructure and friction-wear performances of laser-cladded nano-CeO2–reinforced NiCoCrAlY coatings at high temperature. Proc IMechE, Part B: J Engineering Manufacture 2020; 234(14): 1695–1706.

22.

Montgomery

DC.

Design and analysis of experiments. 9th ed. Hoboken, NJ: John Wiley & Sons, 2017.

23.

Chatterjee

Mahapatra

Bharadwaj

, et al. Drilling of micro-holes on titanium alloy using pulsed Nd:YAG laser: parametric appraisal and prediction of performance characteristics. Proc IMechE, Part B: J Engineering Manufacture 2019; 233: 1872–1889.

24.

Huang

, et al. Prediction of tool wear width size and optimization of cutting parameters in milling process using novel ANFIS-PSO method. Proc IMechE, Part B: J Engineering Manufacture. Epub ahead of print 11 July 2020. DOI: 10.117754405420935262.