Multi-Objective Optimization Study on Production of AlSi10Mg Alloy by Laser Powder Bed Fusion

Abstract

In additive manufacturing, production parameters play a critical role in the microstructure, mechanical properties, and surface quality of a product. The correct selection of these parameters is of great importance for the success of the production process. In this study, the aim was to improve product quality in the additive manufacturing of an AlSi10Mg alloy. The experiments were conducted using a full factorial design, with a constant layer thickness of 0.04 mm. The production parameters included two laser powers (200 and 275 W), two scanning speeds (800 and 1400 mm/s), and two hatch distances (0.08 and 0.14 mm). The performance properties of the produced parts were evaluated according to the relative density and surface roughness criteria. The TOPSIS (Technique for Order Preference by Similarity to Ideal Solution) method was used to optimize both relative density and surface roughness performances simultaneously. The results revealed that the most suitable production parameters for the additive manufacturing of the AlSi10Mg alloy were 275 W laser power, 0.14 mm hatch distance, and 800 mm/s scan speed.

1. Introduction

Additive manufacturing (AM) represents a major innovation in the production of engineering components, enabling the production of parts with complex geometries and lightweight structures that are difficult to achieve with traditional methods [1]. This technology increases design freedom, reduces material waste, and minimizes the number of joining operations [2]. Additive manufacturing can be applied to a wide range of materials, from pure metals to alloys and composites to polymers [3]. The production of the AlSi10Mg alloy by AM methods has become widespread due to the ease of processing with laser-based systems due to the low melting point and eutectic composition of this alloy [4]. The AlSi10Mg alloy is of great importance in the automotive and aerospace sectors [5]. In this context, the production of AlSi10Mg by the selective laser melting (SLM) method is considered a critical issue in terms of research and development studies. SLM is an additive manufacturing technology that enables the production of high-precision solid parts by melting metal powders using laser energy [6]. The quality of manufactured parts is affected by various factors such as material properties, powder properties, and machine features [7]. Additionally, variables in the manufacturing process—laser power, scan speed, layer thickness, infill pattern, production direction, melt pool dynamics, and thermal properties—can significantly affect the performance and quality of final parts [8].

Studies on the production of the AlSi10Mg alloy by the SLM method have comprehensively investigated the effects of parameters in the production process on material properties. In a study by Praneeth et al., the AlSi10Mg alloy was produced by the SLM technique using a Taguchi experimental design. In the study, the effects of variables such as laser power, layer thickness, scan speed, and hatch distance were investigated. The results revealed that the highest relative density and hardness were obtained at 62.50 J/mm³ volumetric energy relative density, and the surface roughness was at the lowest level at 52.08 J/mm³ energy relative density. It was determined that laser power and scan speed were the parameters with the most significant effect on hardness and surface roughness [5]. Siyampaş and Turgut investigated the effects of different production parameters in order to increase product quality in the additive manufacturing of the AlSi10Mg alloy. In the study, parameters such as laser power, scan speed, hatch distance, and layer thickness were tested at various volumetric energy densities. As a result of the tests performed using the Taguchi L9 orthogonal array experimental design, the effects of the parameters on mechanical properties and surface roughness were evaluated. The findings revealed that very high and very low energy densities worsened mechanical properties and increased surface roughness. The best mechanical properties were obtained at 55.82 J/mm³, while the lowest surface roughness was obtained at 54.46 J/mm³ energy relative density. The study emphasizes the importance of selecting the optimal production parameters and shows that 30.15% improvement was achieved in performance indicators. It was also stated that volumetric energy relative density can be used as an important tool in evaluating product quality [3]. In a study by Rios et al., the effects of processing parameters on AlSi10Mg parts produced using the SLM method were investigated in detail. As a result of the experimental design performed with the Taguchi method, nine different processing parameter combinations were determined, and 45 samples were produced with these combinations. Relativistic densities were measured by Archimedes’ principle and metallographic image processing techniques. The findings revealed that laser power is the parameter with the greatest effect on relative density. The Taguchi model suggested 400 W laser power, 1800 mm/s scan speed, and 0 mm laser focal distance as the optimum parameters. With these parameters, 99.98% relative density was achieved. In addition, the mechanical properties of the optimized parts were found to be significantly superior to those of the non-optimized parts [9]. Nine groups of samples were produced from AlSi10Mg powder with various combinations of laser power and scan speed using SLM technology by Huang et al. The study included relative density, tensile test, XRD, LSCM, SEM, and EDS analyses to evaluate the effects of laser power and scan speed on the relative density, microstructure, and tensile properties of the AlSi10Mg alloy produced by SLM. The results showed that increasing the laser power and decreasing the scan speed to a certain level increased the structural continuity of the samples and significantly reduced the porosity. It was found that in combinations with the same energy relative density, the increase in laser power increased the internal temperature of the liquid pool, causing the powder to melt more homogeneously, which improved the relative density and internal porosity. Specimens produced at 225 W/1625 mm/s parameters in particular showed a 3.6% higher relative density and 16.5% higher tensile strength compared to 135 W/975 mm/s parameters [10]. In a study by Liu and Shi, a combination of full factorial experimentation and response surface methodology was developed in order to optimize both the relative density and tensile strength of the AlSi10Mg alloy produced by SLM. The effects of laser power and scan speed on these properties were discussed in detail in the study. The obtained results revealed that there was not always a positive relationship between relative density and tensile strength and that laser power and scan speed had a strong interaction with these properties. It was determined that the developed mathematical models provided high compatibility and accuracy. The optimum production parameters were determined as 340 W laser power and 1870 mm/s scan speed, the experimental relative density obtained with these parameters was found to be 99.34%, and the tensile strength as 328.31 MPa [11]. In the study of Siyampaş and Turgut, the importance of dimensional accuracy in the additive manufacturing of the AlSi10Mg alloy was emphasized. In the study, the geometric tolerances of the parts produced with twelve different laser power and scan speed combinations were investigated. The results showed that the deviation values increased at high and low laser power/scan speed combinations, and the lowest deviations were obtained at a laser power of 250–310 W and a scan speed of 785–974 mm/s. Using the gray relational analysis method, the most suitable production parameters were determined as 290 W laser power and 911 mm/s scan speed. These parameters provide high dimensional accuracy [12]. In a study by Işık et al., the effects of production parameters on surface roughness and geometric tolerances in the additive manufacturing of the AlSi10Mg alloy were investigated. In the experiments carried out using the Taguchi L27 orthogonal array experimental design, it was determined that increasing the laser power positively affected surface roughness and diameter change, while increasing the hatch distance had negative effects on circularity and centrality. In addition, a negative effect of increasing the scan speed on centrality was determined. The most suitable parameter for surface roughness and diameter change was found to be A1B1C3, for circularity change A3B3C1, and for centrality A3B1C1. An ANOVA showed that the most effective parameters on surface roughness, diameter change, circularity change, and centrality were 53.22% laser power, 62.45% hatch distance, 37.23% scan speed, and 40.41% hatch distance, respectively. According to the gray relationship analysis (GRA) results, the most suitable production parameter was determined as A2B1C3 [13].

Studies in the literature show that adjustments and optimizations made in the production process can have a direct effect on the functionality and aesthetic properties of the final product. The meticulous control of production parameters is the key to the production of high-quality and high-performance parts. Özsoy et al. emphasize that since additive manufacturing is still a developing technology, the optimum processing parameters have not yet been fully determined in the literature [4].

In this study, the aim was to increase the efficiency of material, machine, and design processes by determining the right parameter combinations. For this purpose, the effects of parameters such as laser power, hatch distance, and scan speed on relative density and surface roughness were analyzed with a full factorial experimental design using AlSi10Mg alloys produced by the SLM method. The most suitable parameters for two basic performance properties, such as surface roughness and relative density, were determined by TOPSIS, a multi-criteria decision-making method used to select the best parameter combination among alternatives. This study presents a comprehensive methodology for improving the quality and performance of AlSi10Mg parts produced by the SLM method.

2. Materials and Methods

2.1. Production of Test Samples

AlSi10Mg powder from the ADDVALUE Company (Singapore) was used in the production of test samples. The chemical composition and size distribution of the powder are given in

Table 1. Chemical composition of AlSi10Mg powder alloy.

Element	Percentage by Weight
Si	$10.21$
Mg	$0.38$
Mn	$<0.005$
Cu	$<0.005$
F	$0.1$
Zn	$0.0073$
What	$0.0059$
Ti	$<0.005$
Take it	$Bal.$
Pb	$<0.005$
Mr.	$<0.005$

.

The samples were produced with Concept Laser-M2 CUSING brand 3D printer (Concept Laser, Lichtenfels, Germany) located at Gazi University Additive Manufacturing Technologies Application and Research Center (EKTAM).

The experiments were conducted according to the parameter combinations provided in

Table 2. Experimental parameters and levels.

Factors	Symbol	Unit	Variable Levels
			Low	High
Laser Power	LP	W	200	275
Hatch Distance	HD	mm	0.08	0.14
Scan Speed	SS	mm/s	800	1400

, based on a full factorial experimental design. These parameters were determined based on expert opinions from professionals with experience with the equipment and the relevant literature [5,11,13]. The layer thickness was selected as a constant 0.04 mm, and a scan rotation angle of 67° was applied between each layer. The dimensions of the produced samples and the appearance of the printed parts are presented in Figure 1.

2.2. Relative Density and Surface Roughness Measurement

The average surface roughness was measured using a TESA Technology Rugosurf 20 (TESA Technology, Renens, Switzerland) device. Measurements were made from the top surface of the sample in the production direction (Figure 2). The surface roughness of each sample was calculated using the arithmetic average of the measurement values taken from three different points.

The densities of the samples were measured by the Archimedes method using the Shimadzu Analytical Balance, Shimadzu Corporation, Kyoto, Japan (Figure 3). The relative density values were calculated as the ratio between the measured density of each 3D-printed sample and the theoretical density of the AlSi10Mg alloy (2.65 g/cm³).

The experimental design arrangement and the corresponding relative density and surface roughness values are presented in

Table 3. Full factorial experimental design and performance values.

Experiment No.	Laser Power	Hatch Distance	Scan Speed	Surface Roughness	Relative Density
1	275	0.08	800	5.010667	0.991244
2	275	0.14	800	4.421667	0.972
3	200	0.08	800	11.284	0.966104
4	200	0.14	800	13.388	0.988376
5	275	0.14	1400	10.201	0.994442
6	200	0.08	1400	13.53667	0.977332
7	275	0.08	1400	13.20433	0.992571
8	200	0.14	1400	12.94233	0.946044

. Three factors, each with two levels (low and high), were considered in the design. A total of eight experiments were conducted according to these parameter combinations.

2.3. TOPSIS Method: Best Option Approach in Decision-Making Processes

TOPSIS (Technique for Order Preference by Similarity to Ideal Solution) is an effective method widely used in multi-criteria decision-making processes. This method aims to optimize conflicting performance criteria such as density and surface roughness simultaneously. In the fields of material manufacturing and processing, evaluating these two criteria together is critical for obtaining high-quality and high-performance products.

The application of the TOPSIS method follows specific steps:

• Creating a Decision Matrix: A decision matrix is formed that includes the density and surface roughness values of the alternatives (different production methods or parameter settings). This matrix reflects the measurements of both criteria for each alternative.
• Normalization: The decision matrix is normalized to eliminate the impact of different scales. This process helps make the alternatives comparable.
• Weighting: Weights are assigned to the density and surface roughness criteria based on their importance. For example, in some applications, if surface roughness is more critical, a higher weight may be assigned to this criterion.
• Identifying Ideal and Negative Ideal Solutions: From the normalized data, alternatives with the best (ideal) and worst (negative ideal) performance values are identified. The ideal solution represents the highest density and lowest surface roughness, while the negative ideal solution represents the opposite characteristics.
• Distance Calculation: The distances of each alternative from the ideal and negative ideal solutions are calculated, determining how close the alternatives are to the best solution.
• Relative Proximity Value: Using the calculated distances, the relative proximity of each alternative to the ideal solution is assessed. This step ensures that both density and surface roughness are considered together.
• Ranking: Based on the obtained relative proximity values, the alternatives are ranked, and the one with the highest proximity is identified as the most suitable option in terms of density and surface roughness.

In conclusion, the TOPSIS method provides a comprehensive evaluation by simultaneously optimizing important criteria such as density and surface roughness. This approach systematically streamlines complex decision-making processes, facilitating effective and objective results [14,15].

3. Results and Discussions

Data obtained from experimental combinations were analyzed with a full factorial experimental design in the Minitab V20 program. This design was applied to investigate the effects of laser power, hatch distance, and scan speed on surface roughness and relative density, and the results were evaluated with an ANOVA.

The average main effect plots obtained for surface roughness and relative density are shown in Figure 4 and Figure 5, respectively.

In the production of the AlSi10Mg alloy by the SLM method, parameters such as laser power, hatch distance, laser diameter, and scan speed cause significant changes in Ra values [4]. When Figure 4 is examined, it is seen that the surface roughness decreases with the increase in laser power. This result is consistent with the literature [4,6,16]. Surface roughness increases with increasing scan speed [17]. It is thought that this is due to insufficient time for the powder material to melt [6]. The increase in the hatch distance caused a decrease in surface roughness. In the study of Özsoy et al., it was observed that the surface roughness decreased as the hatch distance increased from 0.10 mm to 0.13 mm, and the lowest roughness was obtained at a distance of 0.13 mm [4].

In Figure 5, when the laser power is increased between 200 W and 275 W, the relative density of the samples increases. In the studies of Dzukey and Yang, it is stated that the laser power was used in the range of 200–280 W. Increasing the laser power between 240 and 280 W resulted in a faster relative density increase compared to the increase in the range of 200 and 240 W. It is stated that the relative density increases because the high laser power provides the energy required to melt the powder bed properly [18]. Huang et al. explain that the increase in laser power increases the amount of energy received by the powder, improves the fluidity of the powder during the melting process, and causes the internal voids to decrease and the relative density to increase. They also state that if the laser power is too high, the powder receives excessive energy and causes it to burn or vaporize, and this process causes the formation of internal voids and the relative density to decrease [19]. According to the graph, the increase in the hatch distance leads to a decrease in the relative density. A similar situation was observed in the studies of Yang et al. [20]. Aboulkhair states that wider scan intervals reduce the overlap between layers, and as a result, unmelted powder accumulates and causes voids to form. It is stated that these voids reduce the relative density [21]. When the scan speed was increased from 800 mm/s to 1400 mm/s, the relative density of the sample decreased. According to Chen et al., this decreasing trend is due to the fact that high scan speeds cause the accumulation of unmelted particles during the solidification phase and an increase in the balling phenomenon [22].

A Pareto graph of the standardized effects of laser power, hatch distance and scan speed and the interaction results of these factors on surface roughness and relative density is given in Figure 6.

In the Pareto graph given in Figure 6, laser power (LP), scan speed (SS), and the interactions of these two factors show significant results on the surface roughness. On the other hand, hatch distance (HD) remained below the Pareto line (12.71) and did not show a significant effect on the surface roughness.

In terms of relative density, laser power (LP), hatch distance (HD), scan speed (SS), and the interactions of these factors remained below the Pareto line and did not have a significant effect on the relative density.

Table 4. ANOVA results of factors and interactions for surface roughness.

Source	DF	Adj SS	Adj MS	F-Value	p-Value	% Contribution
Model	6	98.6132	16.4355	1630.07	0.019	99.99
Linear	3	73.5905	24.5302	2432.89	0.015	74.62
Laser Power (W)	1	41.9223	41.9223	4157.84	0.010	42.51
Hatch Distance (mm)	1	0.5422	0.5422	53.77	0.086	0.55
Scan Speed (mm/s)	1	31.1260	31.1260	3087.07	0.011	31.56
2-Way Interactions	3	25.0226	8.3409	827.25	0.026	25.11
LP * HD	1	3.2538	3.2538	322.71	0.035	3.30
LP * SS	1	18.5014	18.5014	1834.97	0.015	18.50
HD * SS	1	3.2674	3.2674	324.06	0.035	3.31
Mistake	1	0.0101	0.0101			0.01
Total	7	98.6232				100.0

S: 0.100413; R² = 99.99%; R² (adj) = 99.93%; R² (pred) = 99.35%.

and

Table 5. ANOVA results of factors and interactions for density.

Source	DF	Adj SS	Adj MS	F-Value	p-Value	% Contribution
Model	6	0.001266	0.000211	0.30	0.881	64.49
Linear	3	0.000749	0.000250	0.36	0.807	38.15
Laser Power (W)	1	0.000655	0.000655	0.94	0.510	33.38
Hatch Distance (mm)	1	0.000087	0.000087	0.12	0.784	4.43
Scan Speed (mm/s)	1	0.000007	0.000007	0.01	0.938	0.34
2-Way Interactions	3	0.000517	0.000172	0.25	0.862	26.32
LP * HD	1	0.000009	0.000009	0.01	0.929	0.44
LP * SS	1	0.000376	0.000376	0.54	0.597	19.18
HD * SS	1	0.000132	0.000132	0.19	0.739	6.70
Mistake	1	0.000697	0.000697			35.51
Total	7	0.001963				100.0

S: 0.0264016; R² = 64.49%; R² (adj) = 0.00%; R² (pred) = 0.00%.

present the factors’ efficiencies and percentage contribution values according to the ANOVA results.

Laser power and scan speed were determined as the parameters with the highest contribution to surface roughness. The regression coefficients obtained for surface roughness are as follows: R² = 0.9999, corrected R² = 0.9993, and estimated R² = 0.9935. The fact that these coefficients are close to 1 shows that the explanatory power of the model is quite high, and it successfully explains the surface roughness. According to these results, the equation of the regression model obtained is as follows:

Surface roughness = 30.63 − 0.14738 Laser Power (W) + 204.07 Hatch Distance (mm) − 0.017719 Scan Speed (mm/s) − 0.5669 Laser Power (W) * Hatch Distance (mm) + 0.000135 Laser Power (W) * Scan Speed (mm/s) − 0.07101 Hatch Distance (mm) * Scan Speed (mm/s)

According to the ANOVA results in Table 5, the activities of the factors did not show a significant and meaningful effect on their binary interactions (p > 0.05). The regression coefficient for relative density was obtained as R² = 0.6449. According to these results, the equation of the regression model obtained is as follows:

Relative density = 1.017 − 0.00033 Laser Power (W) + 0.61 Hatch Distance (mm) − 0.000098 Scan Speed (mm/s) − 0.00093 Laser Power (W) * Hatch Distance (mm) + 0.000001 Laser Power (W) * Scan Speed (mm/s) − 0.00045 Hatch Distance (mm) * Scan Speed (mm/s)

The results obtained with the factorial experimental design were analyzed with the graphs given in Figure 7.

It is thought that multiple responses, such as surface roughness and relative density, should be optimized simultaneously. Improving both properties simultaneously is important in terms of increasing the overall quality and functional adequacy of the product. In order to achieve this goal, the parameters that optimize both responses were determined using the TOPSIS method. This method analyzes the necessary parameters in an integrated manner to maximize relative density while minimizing surface roughness. It is aimed at improving both important properties of the production process simultaneously. The formulas required in the TOPSIS method were used from the work of Alp et al. [14]. The values calculated in the use of the TOPSIS method are given in

Table 6. Values calculated with TOPSIS method.

Normalized Data		Weighted Normalized Data		Si+	Si-	Ci +	Rank
Ra	Rd	Ra	Rd
0.160028	0.358107	0.080014	0.179054	0.009423	0.136394	0.9354	2
0.141217	0.351155	0.070609	0.175577	0.004054	0.145631	0.9729	1
0.360383	0.349025	0.180192	0.174512	0.109703	0.036154	0.2479	4
0.42758	0.357071	0.21379	0.178535	0.143186	0.008007	0.0530	7
0.325795	0.359262	0.162897	0.179631	0.092289	0.053979	0.3690	3
0.432328	0.353081	0.216164	0.176541	0.145588	0.005652	0.0374	8
0.421714	0.358586	0.210857	0.179293	0.140249	0.00994	0.0662	5
0.413346	0.341778	0.206673	0.170889	0.136345	0.009491	0.0651	6

. The weight values for surface roughness and relative density were determined as 0.5 for each. Positive and negative ideal solutions for each response are given in

Table 7. Positive and negative ideal solutions.

	Ra	Rd
Positive ideal	0.070609	0.179631
Negative ideal	0.216164	0.170889

. When Table 6 is examined, it is seen that the closest value to the ideal solution is 0.9729, and the optimum result is obtained in experiment number 2.

4. Conclusions

In this study, in the production of the AlSi10Mg alloy by the Laser Powder Bed Fusion method, the effects of the laser power, hatch distance, and scan speed parameters on surface roughness and relative density at different levels were analyzed using a full factorial experimental design and the TOPSIS method. The results obtained are presented below. As a result of the analyses carried out with the full factorial experimental design, it was found that the surface roughness decreases with increasing laser power (W) and hatch distance (mm) but increases with increasing scan speed (mm/s). In addition, it was observed that the relative density increases with increasing laser power (W) but decreases as the hatch distance (mm) and scan speed (mm/s) increase. As a result of TOPSIS analysis, it was determined that the most suitable experimental conditions were obtained with a combination of 275 W laser power, 0.14 mm hatch distance, and 800 mm/s scan speed. These parameters determined the surface roughness. According to these parameters, the surface roughness was 4.421667 µm, and the relative density was 0.972.

Future research could focus on various areas to deepen the understanding of AlSi10Mg alloy production through the LPBF method and enhance efficiency in applications. This includes investigating different alloys, exploring the effects of preheating, and systematically evaluating the impact of powder particle size on surface roughness and density. Additionally, analyzing the long-term mechanical properties of parts produced by LPBF, utilizing digital twin technology for optimizing production processes, and conducting comparative analyses of different processing combinations and parameters are also significant considerations. Such studies will provide valuable insights that can significantly contribute to improving the effectiveness of applications.