International Journal of Cast Metals Research

ENHANCING ER5356 ALUMINUM ALLOY PERFORMANCE USING COLD METAL

TRANSFER AND WIRE ARC ADDITIVE MANUFACTURING: A MACHINE LEARNING
AND STATISTICAL APPROACH
--Manuscript Draft--

Manuscript Number: IJC2554

Full Title: ENHANCING ER5356 ALUMINUM ALLOY PERFORMANCE USING COLD METAL
TRANSFER AND WIRE ARC ADDITIVE MANUFACTURING: A MACHINE LEARNING
AND STATISTICAL APPROACH

Article Type: Research Article

Keywords: Wire arc additive manufacturing; ER5356; response Surface Methodology; Artificial
Neural Networks; Bed Geometry; Mechanical properties; Microstructure analysis

Abstract: This research focuses on maximizing bead width (BW) and bead height (BH) in Wire
Arc Additive Manufacturing (WAAM) by optimizing welding parameters current, voltage,
and traverse speed in the Cold Metal Transfer process. Using Response Surface
Methodology and artificial neural networks, predictive models were developed,
achieving R² values of 0.964 for BW and 0.9713 for BH. Optimal parameters (current:
135 A, voltage: 16 V, traverse speed: 40 cm/min) yielded a bead width of 5.8 mm and
bead height of 3.65 mm. Microstructural analysis via X-ray diffraction, Scanning
Electron Microscopy, and Electron Backscatter Diffraction revealed significant
variations in crystallographic orientations and micro-cracks. Mechanical testing showed
the bottom section had the highest ultimate tensile stress (294.11 MPa) and yield
strength (190.38 MPa), while the middle section had the highest hardness (74 HV).
This study highlights the importance of optimizing WAAM parameters for improved
mechanical properties and microstructural integrity in additive manufacturing.

Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation
Manuscript - with author details Click here to access/download;Manuscript - with author
details;[Link]

ENHANCING ER5356 ALUMINUM ALLOY PERFORMANCE USING COLD METAL

TRANSFER AND WIRE ARC ADDITIVE MANUFACTURING: A MACHINE
LEARNING AND STATISTICAL APPROACH

Manikandan N* and Mathivanan Arumugam

Department of Mechanical Engineering, SRM Institute of Science and Technology, Ramapuram,

Chennai - 600089

*Corresponding Author: Manikandan N – manisrm2011@[Link]

Abstract

This research focuses on maximizing bead width (BW) and bead height (BH) in Wire Arc
Additive Manufacturing (WAAM) by optimizing welding parameters current, voltage, and
traverse speed in the Cold Metal Transfer process. Using Response Surface Methodology and
artificial neural networks, predictive models were developed, achieving R² values of 0.964 for
BW and 0.9713 for BH. Optimal parameters (current: 135 A, voltage: 16 V, traverse speed: 40
cm/min) yielded a bead width of 5.8 mm and bead height of 3.65 mm. Microstructural analysis
via X-ray diffraction, Scanning Electron Microscopy, and Electron Backscatter Diffraction
revealed significant variations in crystallographic orientations and micro-cracks. Mechanical
testing showed the bottom section had the highest ultimate tensile stress (294.11 MPa) and yield
strength (190.38 MPa), while the middle section had the highest hardness (74 HV). This study
highlights the importance of optimizing WAAM parameters for improved mechanical properties
and microstructural integrity in additive manufacturing.

Keywords: Wire Arc Additive Manufacturing, GMAW, Cold Metal Transfer, Al5356, Response
Surface Methodology, Bead Geometry, Artificial Neural Networks, Mechanical, Microstructural
Analysis.

1 Introduction

Producing big, intricate metal components with great material efficiency is made possible
by Wire Arc Additive Manufacturing (WAAM), which has completely changed metal additive
manufacturing [1–3]. By melting metal wire with an electric arc, WAAM creates layers of
complex shapes that are difficult to produce using conventional manufacturing techniques [4–6].
High deposition rates are achieved and material waste is decreased by this approach [7–9]. The
capacity of Cold Metal Transfer (CMT), a Gas Metal Arc Welding (GMAW) variant, to generate
high-quality welds with less heat input has drawn much interest in WAAM [10,11]. CMT's
precise control over wire feed and arc length results in a stable arc and reduced spatter, crucial
for maintaining the integrity of the deposited material and minimizing distortion [12–14]. Recent
advancements in CMT-GMAW for WAAM include real-time process monitoring, adaptive
control systems, and advanced cooling techniques, which have enhanced process stability and
material properties [15–17]. Despite these advancements, challenges remain in achieving
consistent bead geometry, heat input control, and uniform microstructure in CMT-WAAM [18].
To solve these problems and improve mechanical qualities, welding parameters including
current, voltage, and traverse speed must be optimized. Artificial neural networks (ANNs), in
particular, have demonstrated considerable promise in machine learning models for predicting
and optimizing these welding parameters [19]. ANNs can learn complex relationships between
input parameters and output responses from experimental data, offering accurate predictions and
reducing the inefficiencies of traditional trial-and-error approaches [20]. This predictive
capability helps identify optimal parameters, enhancing bead geometry and overall process
performance, thereby improving the reliability and consistency of fabricated components
[21,22].

Due to its superior strength-to-weight ratio, outstanding weldability, and exceptional

corrosion resistance, the aluminum-magnesium alloy ER5356 is frequently utilized in WAAM
[23,24]. Maintaining the mechanical and microstructural integrity of this material is essential for
high-performance applications since it is widely used in the aerospace, automotive, and marine
sectors [25–27]. In order to improve the bead shape of the aluminum alloy ER5356, this study
focuses on forecasting and optimizing the current, voltage, and traverse speed in the CMT-
WAAM process. ANNs are first employed to forecast the geometry of beads, which serves as a
foundation for additional optimization. The next step involves using Response Surface
Methodology (RSM) to determine the ideal welding settings. The mechanical and
microstructural parameters of the manufactured ER5356 material are examined after the
optimized CMT-WAAM process is executed. This research intends to enhance the mechanical
characteristics and microstructural integrity of ER5356 components by combining parameter
optimization with predictive modeling. This will yield important insights for future WAAM
applications in high-performance sectors.

The literature survey compiles recent advancements in additive manufacturing (AM) of

aluminum alloy 5356, focusing on various techniques and their outcomes. Jiangang et al. [28]
and Kotlarski et al. [29] delve into Gas Metal Arc Welding (GMAW) modes, comparing pulsed
(microMIG) and non-pulsed (microMIG-CC) approaches for thin-wall ER5356 aluminum alloy.
They find that microMIG's rapid cooling enhances grain refinement and reduces porosity,
thereby boosting tensile strength compared to microMIG-CC. Despite microMIG-CC showing
higher hardness due to dislocation aggregation, WAAM-fabricated 5356 aluminum alloy exhibits
superior mechanical properties over as-cast counterparts. Wang et al. [30] explore WAAM of
aluminum alloy 5356, highlighting its feasibility and superior properties compared to cast
aluminum-magnesium alloys. They observe variations in heat dissipation across different wall
regions affecting grain structure and mechanical properties, with the bottom region showing
higher hardness but lower toughness due to interlayer defects. This study underscores the critical
role of controlled heat input in optimizing alloy microstructure and mechanical characteristics,
offering insights for enhancing cooling rates in AM processes.

Aldalur et al. [31] compare three GMAW modes pulsed-GMAW, cold arc, and pulsed-
ACwhile conducting more research on WAAM for aluminum alloy 5356. According to their
research, pulsed-AC mode is a viable alternative for high-quality WAAM with this alloy since it
greatly lowers porosity. Using the Fronius cold metal transfer system, Wieczorowski et al. [32]
examine the porosity in WAAM-produced AA5356 and investigate how route strategy, cooling
duration, and travel speed affect pore properties. Higher travel speeds, they discover, somewhat
decrease pore diameter and overall volume. Computed tomography indicates that porosity has
little effect on mechanical attributes such as yield strength and ultimate tensile strength. Using
the CMT approach, Zhang Z et al. [33] examined the stability and droplet transfer behavior of
the swing arc additive manufacturing process for AZ91 magnesium alloy. The scientists
discovered that variations in arc shape during deposition produce a horizontal force that impacts
droplet stability by looking at electrical waveforms and high-speed droplet photos. They
developed a relationship model that connected the CMT parameters to the Vilarinho regularity
index (IVSC), and they used a multiple-response desirability function technique to optimize
these parameters. Because the optimization could forecast and adjust CMT parameters with
accuracy, process stability was increased.

Gierth et al [34] use gas metal arc welding (GMAW) to examine the WAAM of
AlMg5Mn alloy. It investigates the relationships between temperature-time regimes and energy
input, concentrating on how these relationships affect mechanical characteristics, weld seam
abnormalities, and microstructure. By employing CMT processes, the study achieved controlled
layer geometry and reduced porosity, enhancing tensile strength and material hardness. The
thermal dynamics of the aluminum alloy AA5087 during WAAM is investigated by Sahu M et
al. [35] utilizing both cycle-step (CMT-CS) and traditional CMT modes. The study utilized a
Fanuc Arc Mate 1000iC robot, a Fronius TransPulse Synergic 3200 CMT power source, Ahlborn
Almemo 5690-2 thermocouples to record thermal cycles, and a FLIR E95 camera to investigate
thermal fields. The study reveals notable temperature fluctuations that are impacted by the
selection of arc mode in the additive manufacturing procedure. Mai DS et al [36] focus on
WAAM of thin-walled 308L stainless steel components, optimizing welding parameters for
single weld beads and wall deposition to achieve desired geometries. Their microstructural
analysis identifies dendritic austenite with residual ferrite in grain boundaries, alongside
microhardness measurements reflecting suitable mechanical properties comparable to wrought
308L stainless steel. These findings underscore the potential of WAAM-produced 308L steel for
various industrial applications, validating its mechanical integrity and performance.

The literature survey highlights significant advancements in the additive manufacturing

(AM) of aluminum alloy 5356 but also reveals notable gaps. Specifically, there is a lack of
comprehensive studies on the application of CMT in WAAM for Al5356. Additionally, there is
an absence of research utilizing machine learning techniques to predict and optimize the input
parameters for the CMT-WAAM process of this alloy. This research aims to address these gaps
by pioneering the use of CMT-WAAM for Al5356, integrating ANNs for predictive modeling,
and employing RSM for optimization. This novel approach will enhance process control and
material properties, making it a significant contribution to the field. By combining ANN and
RSM, your study will establish a robust methodology for predicting and optimizing input
parameters, ensuring improved stability, reduced defects, and superior mechanical properties in
the final product.
 The innovation in this research lies in its interdisciplinary approach, merging advanced
machine learning techniques with state-of-the-art additive manufacturing processes. This not
only fills the existing research gaps but also sets a new standard for future studies in AM. The
comprehensive analysis of mechanical and microstructural properties of the optimized WAAM-
produced Al5356 components will provide valuable insights, directly applicable to industrial
applications. The future scope of this research is vast, offering potential advancements in process
efficiency and product quality in various industries that utilize aluminum alloys. By optimizing
the CMT-WAAM process parameters, your research can lead to enhanced tensile strength,
refined microstructures, and reduced porosity, thereby meeting the stringent requirements of
high-performance applications. Expected outcomes include the identification of optimal process
parameters for CMT-WAAM of Al5356, resulting in components with superior mechanical
properties and microstructural characteristics. This work will advance academic understanding
and provide practical solutions for the manufacturing industry, promoting the adoption of
advanced AM techniques for high-quality production.

2 Materials and Methods

ER5356 aluminum wire with a 1.2 mm diameter was used in this study as the deposition
material. The substrate, Al6061 aluminum plate, is well acclaimed for its easy weldability and
strong mechanical qualities. Important measurements for the experimental setup include the
aluminum plate's 10 mm thickness, 150 mm length, and 100 mm breadth. These metrics are
selected to provide uniform deposition quality and consistent heat dispersion by standardizing
the deposition process's operating parameters. Table 1 lists the precise chemical compositions of
the Al6061 aluminum plate and the ER5356 aluminum wire.

Table 1 ER5356 and Al6061 Chemical Composition (wt.%) [37]

Composition Mg Si Fe Cu Mn Cr Ti Al
Al5356 4.5 – 0.05 – 0.05 – 0.06 –
< 0.25 < 0.40 < 0.1 Balance
Wire 5.5 0.20 0.20 0.20

Al6061 0.15- 0.05- 0.04-

0.8-1.2 0.4-0.8 0.7 0.15 Balance
Substrate 0.4 0.2 0.35
 The welding heat for the CMT-WAAM process came from a Panasonic YD-350GR3
power source. The source supports the standard droplet transfer mode and employs the CMT
method. The automatic wire feed speed management in response to the welding current is one of
its key features. The welding current directly relates to an increase in wire feed speed. The
quality and integrity of welded material are preserved by maintaining a stable arc and a
consistent deposition rate, both of which are achieved by this automatic adjustment. The CMT
welding flame was moved using a Panasonic TA1400 6-axis robot, as shown in Figure 1(a),
following the preset deposition routes. This robotic system offers high precision and flexibility,
capable of moving the welding torch in six different directions. During welding, the shielding
gas was 99.99% pure argon. Argon gas was supplied at a rate of 15 liters per minute (L/min),
ensuring a steady and appropriate protective environment surrounding the weld region.
In light of the significance of weld bead quality and geometry in CMT-WAAM thin-
walled component production, the initial part of this work examined the effects of welding
voltage, current, and travel speed on weld bead geometry [36]. Determining the ideal settings to
create a single weld bead with the required form and quality was the aim. Based on experimental
data, the response surface methodology (RSM) and analysis of variance (ANOVA) were used to
compute the bead height (BH) and bead width (BW). Table 2 presents the three selected levels
for every element.
 Figure 1 (a) Experiment setup (b) Single weld (c) Final fabricated wall

Table 2 Different levels of Input parameters

Input Parameters Level 1 Level 2 Level 3

Current (A) 60 120 180
Voltage (V) 12 14 16
Traverse Speed
40 80 120
(cm/min)

The parameter limits were based on the wire manufacturer's recommendations and
verified through several trial runs to ensure weldability and prevent experimental interruptions.

Figure 1(b) shows the 27 trial runs of single weld beads, each approximately 150 mm in
length. Measurements of BH and BW were taken at three positions within the steady region of
each weld bead using a digital caliper (Figure 2) [38], and the average values were reported in
Table 3.
 Figure 2 Determination of Bed width (BW) and Bed Height (BH)

In the second step, the mechanical properties and microstructures of CMT-WAAM

Al5356 walls were investigated to assess the process stability. Using the optimum parameters
identified in the first step, two single-bead multi-layer walls were built. To provide for cooling
and heat dispersion, these walls were constructed using an alternating direction method with a
40-second rest period between layers. There was around 12 mm separating the welding torch's
contact tip and the workpiece.

The walls were fabricated using the CMT-WAAM process. Each wall was constructed
from 16 individual passes, resulting in a final length of 150 mm and a height of approximately 50
mm, as illustrated in Figure 1(c). Following the fabrication, the wire-cut Electrical Discharge
Machining (EDM) method was employed to precisely cut the walls into specimens tailored for
subsequent testing. This precise cutting technique ensured that the specimens met the required
dimensions for various analyses, including microstructure analysis, X-ray diffraction (XRD),
hardness testing, and energy-dispersive spectroscopy (EDS), as demonstrated in Figure 3. The
EDM process facilitated the creation of test samples with high dimensional accuracy, crucial for
reliable tensile and hardness.
 Figure 3 Specimen for Tensile and Hardness Test

3 Results and Discussion

3.1 Artificial Neural Network Prediction

The Python code executed on the Google Colab platform was used to predict bed
geometry, specifically bed width (BW) and bed height (BH), in GMAW-CMT wire arc additive
manufacturing of Al5356 materials using artificial neural networks (ANNs). Initially, the data is
loaded (Table 3) from a CSV file stored in Google Drive. The dataset includes input features
(current, voltage, and travel speed) and target outputs (BW and BH). The features and targets are
extracted and converted into NumPy arrays for ease of manipulation. Rows containing any
missing values (NaNs) are removed to ensure data quality. The characteristics are then
normalized to have a zero mean and unit variance with StandardScaler. Normalization is critical
for enhancing neural network performance since it guarantees that all input characteristics have
the same scale. Next, the normalized data is divided into training and validation sets. This is
accomplished using an 80-20 split, which ensures that 20% of the data is retained for validation.
Separate divisions are used for the BW and BH objectives. The training set trains the models,
whereas the validation set evaluates their performance. Two separate ANN models are defined,
one for predicting BW and the other for predicting BH. Both models use a sequential architecture
with dense (fully connected) layers. The BW model consists of three hidden layers with 4, 3, and
3 neurons using the ReLU activation function. There is only one neuron in the output layer, and
it does not have an activation function, hence the output is continuous. With two hidden layers of
four neurons each and a comparable output layer, the BH model's design is a little bit simpler.
Because it may create non-linearity and help the network learn complicated patterns, the ReLU
activation function was used (Figure 4).

The Adam optimizer is used to build the models, and the loss function is mean absolute
error or MAE. The variable learning rate features and efficiency of the Adam optimizer led to its
selection. Small batch sizes (5 for BW and 6 for BH) are used to train both models throughout
1000 epochs. In order to reduce loss, the training approach involves repeatedly iterating over the
training data and adjusting the model weights. For further research, the training history is
preserved, together with the validation and training losses for every epoch.

(a) (b)

Figure 4 ANN Architecture (a) Bed Width (b) Bed Height Model
3.1.1 Model Performance

Figure 6 displays the connection between the actual and predicted values as well as the
training and validation losses throughout epochs. The performance graphs for both models make
it clear that the training process was effective. Figure 5(a) displays the training and validation
loss for the bed width prediction model across 1000 epochs. Both the training and validation
losses dramatically decrease in the first few epochs and stabilize around a low value, indicating
effective learning without overfitting or underfitting. The model's ability to effectively generalize
to new data is demonstrated by the training and validation loss curves' tight alignment with the
high R² value of 0.964 and the low MAE of 0.1123, which indicate great predictive performance.

Figure 5 ANN Performances for both BW and BH Models

When the projected and actual bed width values are compared in Figure 5(c), it can be
seen that a perfect match is represented by the blue data points that are near the red line. The
high R² value of 0.964 and the closeness of the points to the line imply that the model predicts
bed width with accuracy. It is further confirmed by the low MAE of 0.1123 that the average
prediction error is modest. Comparably, the training and validation loss for the bed height
prediction model across 1000 epochs is displayed in figure 5(b). Effective learning is shown by
both the training and validation losses rapidly decreasing and stabilizing. The close alignment of
training and validation losses indicates good generalization to the validation data, consistent with
the high R² value of 0.9713 and the low MAE of 0.0952, demonstrating excellent model
performance.

The actual and anticipated values for bed height are contrasted in Figure 5(d). The red
line is surrounded by a small cluster of blue data points, suggesting good forecast accuracy. The
model's high R² value of 0.9713 and the close clustering of points around the perfect fit line
demonstrate how well it predicts bed height. An average prediction error that is modest is shown
by the low MAE of 0.0952. The ANN models are capable of accurately predicting bed geometry
in GMAW CMT wire arc additive manufacturing of Al5356 materials, as shown by the
performance plots and metrics. The high R² values and low MAE values reflect the models' high
predictive accuracy, making them reliable tools for predicting bed width and bed height based on
input parameters such as current, voltage, and travel speed.

3.2 Response surface methodology

To optimize BW and BH in the CMT-WAAM process, welding parameters such as

current, voltage, and traverse speed were systematically adjusted using Response Surface
Methodology (RSM) [39,40]. This study utilized Design Expert 13 software to investigate the
effects of these parameters on weld bead geometry. RSM facilitated the development of
predictive models and the identification of optimal conditions, ensuring high-quality aluminum
components [41,42]. This approach enhanced the production process by providing valuable
insights into the relationships between welding settings and the final bead characteristics, thereby
improving accuracy and efficiency.
 Table 3 Bead width and Height for 27 welds

Voltage Traverse Speed BW BH

[Link] Current (A)
(V) cm/min (mm) (mm)
1. 60 12 40 4.25 3.16
2. 60 12 80 3.78 2.88
3. 60 12 120 3.51 2.45
4. 60 14 40 4.68 2.91
5. 60 14 80 4.21 2.68
6. 60 14 120 3.96 2.21
7. 60 16 40 5.05 2.63
8. 60 16 80 4.74 2.31
9. 60 16 120 4.32 1.98
10. 120 12 40 4.86 4.12
11. 120 12 80 4.35 3.98
12. 120 12 120 4.18 3.54
13. 120 14 40 5.47 3.84
14. 120 14 80 4.98 3.62
15. 120 14 120 4.53 3.23
16. 120 16 40 5.86 3.55
17. 120 16 80 5.41 3.21
18. 120 16 120 4.88 2.96
19. 180 12 40 4.72 3.54
20. 180 12 80 4.22 3.38
21. 180 12 120 3.85 2.96
22. 180 14 40 5.21 3.43
23. 180 14 80 4.77 3.11
24. 180 14 120 4.42 2.75
25. 180 16 40 5.56 3.18
26. 180 16 80 5.38 2.85
27. 180 16 120 4.77 2.51

Quadratic models were chosen for bead height (BH) and bead width (BW) following a
comprehensive statistical significance analysis and observation of the great consistency between
the adjusted and projected R² values. Equations (1) and (2) in this study show how current (I),
voltage (V), and traversal speed (S) affect the special performance characteristics of the CMT-
WAAM process. The high coefficients of correlation (R²) values, 0.9881 for BW and 0.9923 for
BH confirm the models' predictive accuracy, demonstrating a comprehensive and precise
understanding of the process parameters.

BW = -1.56074+0.02969*I+0.516944*V-0.008132*S+0.000243*I*V-9.02778E-06*I*S-
0.000219*V*S-0.000118*I2-0.010694*V2+0.000014*S2 (1)

BW = +1.72130+0.050384*I-0.006111*V-0.001917*S++0.00016*I*V+4.51389E-06*I*S-
0.000042*V*S-0.000203*I2-0.005139*V2-0.000038*S2 (2)

3.2.1 Analysis of Variance (ANOVA)

The ANOVA table (Table 4) for analyzing bead width (BW) in the GMAW-WAAM
process for ER5356 aluminum indicates significant influences from the process parameters. The
model shows a high F-value (156.59) and a very low p-value (< 0.0001), indicating its overall
significance. Voltage (B) has the highest contribution (43%) to the variation in bed width,
followed by Traverse Speed (C) at 33% and Current (A) at 12%. Therefore, Voltage (B) is
identified as the most influential parameter for bed width. Interaction terms (AB, AC, BC) and
some quadratic terms (B², C²) show no significant influence, as evidenced by their high p-values
and low contributions. The high R² values (adjusted R² = 0.9818, predicted R² = 0.9689) confirm
the model’s accuracy in predicting bead width based on these parameters.

Table 4 ANOVA for BW Model

Sum of Mean
Source df F-value p-value Contribution
Squares Square
Model 8.88 9 0.987 156.59 < 0.0001 significant
A-Current 1.08 1 1.08 170.63 < 0.0001 significant 12%

B-Voltage 3.78 1 3.78 599.88 < 0.0001 significant 43%

C-Traverse
2.91 1 2.91 461.99 < 0.0001 significant 33%
Speed
AB 0.0102 1 0.0102 1.62 0.2203 0%
AC 0.0056 1 0.0056 0.8937 0.3577 0%
BC 0.0037 1 0.0037 0.583 0.4556 0%
A² 1.08 1 1.08 171.48 < 0.0001 significant 12%
B² 0.011 1 0.011 1.74 0.2044 0%
C² 0.003 1 0.003 0.4701 0.5022 0%
Residual 0.1072 17 0.0063
Cor Total 8.99 26
Adjusted R² 0.9818
Predicted R² 0.9689

The ANOVA table (Table 5) for bead height (BH) in the CMT-WAAM process for
ER5356 aluminum demonstrates significant contributions from several process parameters. The
model is highly significant, with an F-value of 243.48 and a p-value of less than 0.0001. Among
these factors, Traverse Speed (C) has the highest contribution (25%) to the variation in bead
height, followed by Voltage (B) at 17% and Current (A) at 15%. This suggests that Traverse
Speed (C) has the most influential impact on bead height in the CMT-WAAM process for
ER5356 aluminum. The high adjusted R² (0.9882) and predicted R² (0.9799) values confirm the
model's strong predictive accuracy.

Table 5 ANOVA for BH Model

Sum of Mean
Source df F-value p-value Contribution
Squares Square
Model 7.52 9 0.8353 243.48 < 0.0001 significant

A-Current 1.13 1 1.13 327.91 < 0.0001 significant 15%

B-Voltage 1.3 1 1.3 377.77 < 0.0001 significant 17%
C-Traverse
1.85 1 1.85 539.11 < 0.0001 significant 25%
Speed
AB 0.0044 1 0.0044 1.28 0.2727 0%
AC 0.0014 1 0.0014 0.4105 0.5303 0%
BC 0.0001 1 0.0001 0.0389 0.8461 0%
A² 3.22 1 3.22 937.64 < 0.0001 significant 43%
B² 0.0025 1 0.0025 0.7389 0.402 0%
C² 0.022 1 0.022 6.41 0.0215 0%
Residual 0.0583 17 0.0034
Cor Total 7.58 26
Adjusted R² 0.9882
Predicted R² 0.9799

3.2.2 The influence of processing parameters on the geometry of weld beads

 Figures 6 & 7 consist of response surface plots (a-c) and a histogram (d) that illustrate the
influence of different combinations of input parameters current (A), voltage (V), and traverse
speed (cm/min) on BW and BH in wire arc additive manufacturing using ER5356 wire. Figure
6(a) shows how voltage (V) and current (A) jointly influence the bead width (BW). As voltage
increases from 12V to 16V, the bead width initially increases, reaches a peak, and then slightly
decreases. This indicates an optimal voltage range of around 14-15V for maximizing bead width.
Similarly, as the current increases from 60A to 180A, the bead width initially increases and then
slightly decreases, with an optimal current range around 120-150A. The surface plot peaks at a
region where both voltage and current are moderately high, suggesting that a balance between
these two parameters is crucial for achieving maximum bead width.
 Figure 6 Influence on Input parameters in BW model (a-c) Response surface plots (d)
Histogram

Figure 6(b) depicts the combined effect of current (A) and traverse speed (cm/min) on
bead width (BW). As current increases, bead width initially increases and then decreases, with an
optimal range around 120-150A. Conversely, as traverse speed increases from 40 cm/min to 120
cm/min, bead width consistently decreases. Lower traverse speeds lead to wider beads due to
more prolonged exposure to heat. Therefore, the optimal bead width is achieved at higher
currents and lower traverse speeds, suggesting that more heat input and slower movement
enhance material deposition. Figure 6(c) shows the interaction between voltage (V) and traverse
speed (cm/min) on bead width (BW). Increasing voltage from 12V to 16V initially results in
wider beads, reaching a maximum of around 14-15V, and then slightly decreasing. Similar to
previous observations, increasing traverse speed reduces bead width. Lower speeds (~40
cm/min) yield the widest beads. The surface plot indicates that the widest beads are obtained at a
moderate voltage (around 14-15V) and low traverse speed (around 40 cm/min). Figure 6(d)
displays the frequency distribution of bead width measurements. The histogram appears to be
moderately distributed with a peak frequency around the mid-range bead width values (around 5
mm). This suggests that the majority of bead widths fall within this range, reflecting the
consistency of the process parameters. This frequency distribution implies that the selected
parameter ranges effectively produce desirable bead widths, with fewer instances of extreme
values.

Figure 7(a) shows the combined effect of voltage (V) and current (A) on bead height
(BH). As voltage increases from 12V to 16V, bead height decreases. Lower voltages (~12V)
result in higher bead heights because they concentrate the heat, resulting in deeper penetration.
Higher voltages spread the heat over a larger area, reducing penetration and bead height. Bead
height increases with current up to around 120A, after which it decreases slightly. Moderate
currents (around 120A) provide sufficient heat input for adequate melting and deposition without
causing excessive spatter or overheating, which can reduce bead height. The surface plot peaks at
lower voltages and moderate currents, suggesting these conditions are optimal for maximizing
bead height.
 Figure 7 Influence on Input parameters in BH model (a-c) Response surface plots (d)
Histogram

Figure 7(b) depicts the combined effect of current (A) and traverse speed (cm/min) on bead
height (BH). Similar to Figure 7(a), bead height increases with current up to about 120A and
then decreases. As traverse speed increases from 40 cm/min to 120 cm/min, bead height
decreases. Lower speeds (~40 cm/min) result in higher beads due to longer heat exposure. Lower
traverse speeds allow more time for heat to be deposited in one location, increasing bead height,
while higher speeds reduce the heat input per unit length, leading to shorter beads. Optimal bead
height is achieved at moderate currents and lower traverse speeds, indicating a balance between
heat input and deposition time is crucial. Figure 7(c) shows the interaction between voltage (V)
and traverse speed (cm/min) on bead height (BH). Bead height decreases with increasing
voltage. Lower voltages (~12V) yield higher beads. As seen in previous plots, increasing traverse
speed reduces bead height. Lower speeds (~40 cm/min) lead to higher beads. The surface plot
indicates that the highest bead heights are obtained at low voltages and low traverse speeds.
Figure 7(d) displays the frequency distribution of bead height measurements. The histogram
shows a peak around the mid-range bead height values (~2.8-3.3 mm), suggesting that most bead
heights fall within this range, reflecting consistent process parameters. In summary, for optimal
bead width and height in wire arc additive manufacturing, moderate current (around 120-150A),
low voltage (around 12-15V), and low traverse speed (around 40 cm/min) are recommended.
These conditions ensure adequate heat input and deposition time, leading to desirable bead
characteristics while maintaining process consistency.

3.2.3 Optimization of Input Parameters

Using Design Expert 13 software, the study's main objective was to determine the
best set of experimental variables to maximize response values. The goal of multi-response
optimization was to maximize bead height (BH) and bead width (BW). The study found
configurations that successfully improve the CMT-wire arc additive manufacturing (WAAM)
process's features for improved performance and precision by evaluating and adjusting input
parameters. The ideal parameters for the optimization procedure shown in Figure 8 were
determined to be a traverse speed of 40 cm/min, a voltage of 16 V, and a current of 135 A.
Maximum values of 5.8 mm for BW and 3.65 mm for BH were obtained with this arrangement.
With careful selection and optimization of important process variables, these revised parameters
demonstrate the superior outcomes in BW and BH and underscore the efficacy of diligent fine-
tuning in refining the CMT-WAAM process.
 Figure 8 Confirmation of the location of input parameters
The ideal parameters for the optimization procedure shown in Figure 7 were determined
to be a traverse speed of 40 cm/min, a voltage of 16 V, and a current of 135 A. This
configuration achieved maximum values for bead width (BW) at 5.8 mm and bead height (BH)
at 3.65 mm. These refined parameters highlight the effectiveness of meticulous fine-tuning in
optimizing the WAAM process, demonstrating superior outcomes in BW and BH through
careful selection and optimization of key process variables.

3.3 XRD Analysis

Figure 9 illustrates the X-ray diffraction (XRD) analysis of Al5356 material fabricated
through Cold Metal transfer (CMT) and Wire Arc Additive Manufacturing (WAAM) revealing
significant microstructural variations across the bottom, middle, and top regions of the material.
The identified peaks, corresponding to the (111), (200), (220), and (311) crystallographic planes,
exhibit a consistent trend of decreasing intensity from the bottom to the top [43]. The bottom
sample, represented by the black line, shows the highest peak intensities, indicating a well-
defined crystalline structure likely due to slower cooling rates and more stable thermal
conditions during the fabrication process. The middle sample, indicated by the blue line, shows a
reduction in peak intensity, suggesting a decrease in crystallinity possibly due to varying thermal
cycles that affect grain orientation and structure.

Figure 9 XRD image of Bottom, middle, and top samples

The top sample, represented by the red line, has the lowest peak intensities, indicating the
lowest degree of crystallinity, likely a result of rapid cooling rates leading to finer grains or
partial amorphous structures. Additionally, traces of the oxide phase Al2O3 were detected, likely
due to the sample preparation procedure involving mechanical wet grinding, highlighting
aluminum's sensitivity to oxidation. Despite the possibility of small Al3Mg2 intermetallic phase
content, it was not detected by the diffractometer in this study. This XRD analysis highlights the
microstructural differences within the Al5356 material, providing crucial insights into its
mechanical properties and performance.

3.4 SEM Analysis of Fabricated Samples

The Scanning Electron Microscope (SEM) images (Figure 10) depict the surface
morphology of the bottom, middle, and top samples of Al5356 material fabricated through CMT-
WAAM. Each image was taken using back-scattered electrons (BSE) at a 500× magnification,
allowing for detailed observation of surface features such as micro-cracks, micro-pores, and
undissolved particles. Figure 10(a) shows the SEM image of the bottom sample. The image
reveals the presence of two micro-cracks, each approximately 70 μm in length. These micro-
cracks indicate potential weaknesses in the material's structure, possibly due to the stresses and
thermal gradients experienced during the additive manufacturing process. Additionally, several
micro-pores are visible in the specimen, which could be attributed to gas entrapment or
solidification shrinkage during the welding process. The image also shows undissolved
magnesium (Mg) particles scattered throughout the structure. These particles suggest incomplete
dissolution or segregation of alloying elements during the fabrication process, which could
impact the material's mechanical properties. Figure 10(b) presents the SEM image of the middle
sample.

Figure 10 SEM images of fabricated bottom, middle, and top samples.

 The surface morphology shows a few micro-cracks, with an average length of about 10
μm. The reduced number and size of micro-cracks compared to the bottom sample may indicate
a more stable thermal environment during the deposition of this layer, resulting in fewer stress-
induced defects. Similar to the bottom sample, undissolved Mg particles are observed, indicating
consistent alloying element behavior throughout the middle section of the material. The presence
of these particles suggests that there may be areas within the material where alloying elements
are not fully integrated into the matrix, potentially affecting its uniformity and strength. Figure
10(c) depicts the SEM image of the top sample. The surface exhibits numerous micro-sized
pores with irregular shapes distributed across the sample. These pores could be the result of rapid
cooling rates at the top of the fabricated material, leading to incomplete solidification and gas
entrapment. The presence of undissolved Mg particles is again noted, similar to the other two
samples. The irregular distribution of pores and undissolved particles at the top layer suggests
that the rapid cooling and solidification rates may hinder the complete mixing and integration of
alloying elements, potentially compromising the material's mechanical integrity.

From the overall Analysis, the bottom sample exhibits the longest micro-cracks, while the
middle sample shows shorter and fewer micro-cracks. Micro-pores are present in all samples but
are most prominent and irregularly shaped in the top sample. These pores may result from gas
entrapment or solidification shrinkage, influenced by the cooling rates during the fabrication
process. The SEM analysis underscores the microstructural differences across the bottom,
middle, and top regions of the Al5356 material. These differences are likely due to varying
thermal conditions and cooling rates experienced during the CMT-WAAM process.
Understanding these microstructural characteristics is crucial for optimizing the fabrication
process and improving the material's overall performance and reliability.

3.5 EBSD Analysis

Figure 11 illustrates the Electron Backscatter Diffraction (EBSD) analysis of a Cold

Metal Transfer (CMT) - Wire Arc Additive Manufacturing (WAAM) fabricated Al5356 material
explores the microstructural characteristics and crystallographic texture at three different sections
along the building direction: top, middle, and bottom. The EBSD maps, displayed in the middle
column, use various colors to indicate different crystallographic orientations of the grains. The
right column presents pole figures and inverse pole figures, illustrating the density of specific
crystallographic orientations and providing a detailed breakdown of texture components. The
harmonic texture analysis adds a quantitative measure of texture complexity, enhancing the
understanding of microstructural variations [44,45].

Figure 11 EBSD mapping analysis top, middle and bottom portion

The analysis reveals significant variations in grain structure and orientation along the
building direction. The top section displays a variety of grain orientations with a complex
texture, indicated by high-density regions in the pole figures and a harmonic texture maximum
intensity of 2.760. The middle section shows elongated grains and a pronounced texture with a
maximum intensity of 7.943, while the bottom section has more uniform and elongated grains
with a moderately complex texture and a maximum intensity of 5.623. These variations influence
the material's mechanical properties, with the top section potentially exhibiting different strength
and ductility compared to the more uniform bottom section.

Understanding these microstructural characteristics is crucial for optimizing the CMT-WAAM

process to achieve desired material properties, such as improved mechanical strength and
ductility [46]. This detailed EBSD and harmonic texture analysis provide valuable insights for
predicting the material's behavior under different conditions, essential for designing reliable and
robust components in engineering applications.

3.6 Mechanical Properties

The mechanical characteristics of Al5356 produced via Cold Metal Transfer (CMT) -
Wire Arc Additive Manufacturing (WAAM) are depicted in the graph (Figure 12). Samples were
gathered from the top, middle, bottom, and vertical regions of the produced item, among other
locations. Ultimate Tensile Stress (UTS), Yield Strength (YS), and Elongation (El%) are among
the qualities that are assessed. The mechanical properties vary across different sections of the
fabricated part, which is significant in understanding the material's performance and consistency.
The bottom section exhibits the highest UTS (294.11 MPa) and YS (190.38 MPa), suggesting
that this area experienced the most favorable thermal cycles and solidification conditions, likely
resulting in finer microstructures that enhance strength. The elongation in the bottom section
(18.94%) is also relatively high, indicating a good balance between strength and ductility. The
vertical section shows moderate UTS (256.52 MPa) and YS (147.18 MPa) but still maintains a
decent elongation (15.66%), suggesting a more balanced mechanical performance with
acceptable ductility.

This could be due to more uniform thermal exposure and grain structure in the vertical
build direction, which typically experiences different cooling rates and thermal gradients
compared to horizontal layers. The top and middle sections exhibit intermediate mechanical
properties, with the middle section showing slightly higher values in both UTS and YS compared
to the top. This might be due to the middle section benefiting from more stable thermal
conditions during the additive process, while the top section could be affected by surface cooling
and other environmental factors.
 Figure 12 Mechanical Properties of Fabricated samples.

This research indicates that the mechanical properties of Al5356 can be significantly
improved by optimizing WAAM parameters such as current, voltage, and traverse speed. These
optimizations lead to enhanced UTS, YS, and elongation percentages compared to previous
works and literature. Improved control over the thermal cycles and deposition parameters likely
results in refined microstructures and reduced defects, contributing to the overall enhancement in
mechanical properties. The graph is crucial for understanding the distribution of mechanical
properties within the WAAM-fabricated Al5356 part. It demonstrates the effectiveness of
parameter optimization in improving material performance, providing valuable insights for
future research and practical applications in additive manufacturing. This analysis highlights the
importance of fine-tuning process parameters to achieve desired mechanical properties, which is
vital for the structural integrity and reliability of additively manufactured components.
 Figure 13 Stress-Strain Curve (a) Horizontal Samples (b) vertical samples

The stress-strain curves for the Al5356 samples fabricated using Wire Arc Additive
Manufacturing (WAAM) illustrate the mechanical behavior of the material at different sections:
horizontal top (HT), middle (HM), and bottom (HB). Figure 13(a) shows that the bottom section
(HB) exhibits the highest ultimate tensile stress (UTS) and yield strength (YS), indicating
superior mechanical properties compared to the middle (HM) and top (HT) sections.
Specifically, the bottom section reaches UTS of approximately 294.11 MPa and YS of around
190.38 MPa. The middle section (HM) displays intermediate mechanical properties, with the
highest hardness value recorded at 74 HV. The bottom section (HB) shows the lowest UTS and
YS, suggesting less desirable mechanical characteristics. The stress-strain curves reveal that the
bottom portion (HB) exhibits the highest strength and ductility, followed by the middle (HM),
and then the top (HT). In figure 13(b), the curves represent samples taken from three different
locations (VT1, VT2, and VT3) in vertically oriented specimens. Here, VT1, VT2, and VT3
samples show varying mechanical properties, with VT1 exhibiting the highest strength and
ductility, VT2 showing intermediate properties, and VT3 show the lowest. The inset image
shows the fractured vertical specimens, similarly marked and sized for comparison.
 Figure 14 Hardness value for the horizontal zones

The microhardness analysis (Figure 14) of the WAAM fabricated Al5356 material
reveals the hardness values across three sections along the building direction: top, middle, and
bottom. The bar graph displays the hardness values measured in Vickers Hardness (HV0.5) for
each section. The bottom section has a hardness of 71.8 HV, the middle section has a hardness of
74 HV, and the top section has a hardness of 73.8 HV. This analysis indicates that the middle
section has the highest hardness, followed closely by the top section, with the bottom section
having the lowest hardness. The slight increase in hardness from the bottom to the top can be
attributed to the differences in thermal histories during the WAAM process. As the material is
deposited layer by layer, the bottom section experiences a longer thermal exposure and slower
cooling rates compared to the middle and top sections. This extended thermal exposure can lead
to grain growth and potentially softer material properties. The middle section, with the highest
hardness, might have experienced optimal thermal conditions, balancing cooling rates and
thermal exposure, leading to finer grain structures and increased hardness. The top section,
although slightly lower in hardness than the middle, still shows increased hardness compared to
the bottom section. This could be due to faster cooling rates near the top of the build, resulting in
a finer microstructure. Overall, the microhardness analysis provides valuable insights into the
material's mechanical properties, confirming that the WAAM process can result in variations in
hardness along the building direction. These variations are essential for understanding the
material's performance in different applications and optimizing the WAAM process to achieve
desired mechanical properties throughout the entire build.

3.7 SEM Analysis of Tensile Fracture Surface

The SEM (Scanning Electron Microscope) images (Figure 15) show the tensile test
fracture surfaces of Al5356 samples fabricated using CMT-WAAM. The images are taken from
different sections: Horizontal Bottom (H-B), Horizontal Middle (H-M), Horizontal Top (H-T),
and Vertical (V). These images help to analyze the microstructural features, such as pores and
cracks,that influence the mechanical properties of the material.

The SEM image of the H-Bsection shows limited pores scattered across the fracture
surface. These pores are likely formed due to gas entrapment during the solidification process.
The presence of pores can act as stress concentrators, reducing the material's overall mechanical
strength. The H-M section exhibits both pores and cracks. The pores are similar to those
observed in the H-B section, but the presence of cracks indicates that the material may have
experienced thermal stresses or solidification shrinkage during the WAAM process. These cracks
can propagate under tensile stress, leading to premature failure. The H-T section shows a higher
density of pores and several prominent cracks. The top layers in WAAM processes are more
prone to defects due to rapid cooling and less stable thermal conditions. These factors can lead to
the formation of larger pores and more significant cracking, which detrimentally affects the
material's tensile properties. The V section reveals multiple cracks along with some pores. The
cracks in the vertical direction suggest that anisotropic thermal gradients and residual stresses
during the additive process contribute to crack formation. The vertical build direction typically
experiences different cooling rates compared to horizontal sections, leading to variations in
defect formation.
Figure 15 SEM images of the fracture surface for Horizontal Bottom (H-B), Middle (H-M),
Top (H-T) and Vertical (V)

The SEM images indicate that the fabricated Al5356 material contains microstructural
defects such as pores and cracks, which are critical in determining the tensile properties of the
material. The distribution and density of these defects vary with the position within the
fabricated part, affecting the overall mechanical performance. The SEM analysis of the tensile
test fracture surfaces reveals that the mechanical properties of CMT-WAAM-fabricated Al5356
are influenced by microstructural defects like pores and cracks. These defects are more
pronounced in certain sections due to variations in thermal conditions and solidification behavior
during the additive manufacturing process. Optimizing the CMT-WAAM parameters, such as
current, voltage, and traverse speed, is crucial to minimizing these defects and improving the
material's mechanical performance. Understanding and mitigating the formation of pores and
cracks can lead to the production of more reliable and robust additively manufactured
components.
4 Conclusions

This study highlights the effectiveness of combining artificial neural networks (ANNs)
and Response Surface Methodology (RSM) in optimizing the welding parameters for the
GMAW Cold Metal Transfer (CMT) wire arc additive manufacturing process of Al5356
material. Initially, ANNs were used to predict bead geometry with high accuracy, achieving R²
values of 0.964 for bead width (BW) and 0.9713 for bead height (BH), and low mean absolute
errors (MAE) of 0.1123 and 0.0952, respectively. These predictive models provided a solid
foundation for further optimization using RSM.

By identifying the optimal parameters current of 135 A, voltage of 16 V, and traverse

speed of 40 cm/min, this research achieved a bead width of 5.8 mm and a bead height of 3.65
mm. The XRD and SEM analyses revealed significant microstructural variations across different
sections, with notable crystallographic orientations and micro-cracks. EBSD analysis further
highlighted variations in grain structure and orientation, contributing to the differences in
mechanical properties observed.

The mechanical properties tests showed that the bottom section had the highest ultimate
tensile stress (UTS) of 294.11 MPa and yield strength (YS) of 190.38 MPa, while the middle
section exhibited the highest hardness value of 74 HV. The integration of ANNs and RSM not
only facilitated the optimization process but also ensured the reliability and accuracy of the
predictions. In summary, this research provides comprehensive insights into optimizing CMT-
WAAM parameters to enhance mechanical properties and microstructural integrity. The
successful application of ANNs for predictive modeling, followed by RSM for optimization,
underscores their combined effectiveness in additive manufacturing. Future work can explore the
integration of real-time monitoring and adaptive control systems to further refine the WAAM
process, potentially extending its applications in various high-performance industrial sectors.
ORCID

Corresponding Author: Manikandan N - [Link]

Declaration of Competing Interests

There isn't a conflict of interest related to this study, the authors declare. They add that none of
their observable financial interests or interpersonal connections could be interpreted as affecting
the conclusions made in this study.

Data availability
Python code is available as a supplementary file

Author Contribution:

Manikandan N: Investigation, materials, implementation of the computer code and supporting

algorithms, Validation Writing - Original Draft. Mathivanan Arumugam: Development or
design of methodology; Research Administration.
REFERENCES

[1] Tawfik MM, Nemat-Alla MM, Dewidar MM. Enhancing the properties of aluminum
alloys fabricated using wire + arc additive manufacturing technique - A review. J Mater
Res Technol 2021;13:754–68. [Link]

[2] Geng H, Li J, Xiong J, Lin X, Zhang F. Optimization of wire feed for GTAW based
additive manufacturing. J Mater Process Technol 2017;243:40–7.
[Link]

[3] Vora J, Parikh N, Chaudhari R, Patel VK, Paramar H, Pimenov DY, et al. applied sciences
Optimization of Bead Morphology for GMAW-Based Wire-Arc Metal-Cored Wires
2022:1–18.

[4] Li F, Chen S, Shi J, Tian H, Zhao Y. Evaluation and optimization of a hybrid

manufacturing process combining wire arc additive manufacturing with milling for the
fabrication of stiffened panels. Appl Sci 2017;7. [Link]

[5] Ren L, Gu H, Wang W, Wang S, Li C, Wang Z, et al. Effect of Mg content on

microstructure and properties of Al-Mg alloy produced by the wire arc additive
manufacturing method. Materials (Basel) 2019;12:1–12.
[Link]

[6] Ahsan MRU, Seo G-J, Fan X, Liaw PK, Motaman S, Haase C, et al. Effects of process
parameters on bead shape, microstructure, and mechanical properties in wire + arc
additive manufacturing of Al0.1CoCrFeNi high-entropy alloy. J Manuf Process
2021;68:1314–27. [Link]

[7] Waqas A, Qin X, Xiong J, Wang H, Zheng C. Optimization of process parameters to

improve the effective area of deposition in GMAW-based additive manufacturing and its
mechanical and microstructural analysis. Metals (Basel) 2019;9.
[Link]

[8] Yan Y, Si J, Di X, Guo Y, Han Q, Liu C. Minimizing porosity and optimizing

microstructure of hot-wire arc additive manufactured Al-Cu-Mg-Ag alloy for strength
increment. J Manuf Process 2024;118:89–102.
[Link]
[9] Mohd Mansor MS, Raja S, Yusof F, Muhamad MR, Manurung YH, Adenan MS, et al.
Integrated approach to Wire Arc Additive Manufacturing (WAAM) optimization:
Harnessing the synergy of process parameters and deposition strategies. J Mater Res
Technol 2024;30:2478–99. [Link]

[10] Liu Y, Liu Z, Zhou G, He C, Zhang J. Microstructures and Properties of Al-Mg Alloys
Manufactured by WAAM-CMT. Materials (Basel) 2022;15.
[Link]

[11] Fan S, Guo X, Tang Y, Guo X. Microstructure and Mechanical Properties of Al-Cu-Mg
Alloy Fabricated by Double-Wire CMT Arc Additive Manufacturing. Metals (Basel)
2022;12:416. [Link]

[12] He S, Zhang Z, Li H, Zhang T, Lu X, Kang J. Multi-Objective Optimization for Forming

Quality of Laser and CMT-P Arc Hybrid Additive Manufacturing Aluminum Alloy Using
Response Surface Methodology. Actuators 2024;13. [Link]

[13] Su C, Chen X, Gao C, Wang Y. Effect of heat input on microstructure and mechanical
properties of Al-Mg alloys fabricated by WAAM. Appl Surf Sci 2019;486:431–40.
[Link]

[14] Liu K, Chen X, Zhang Y, Pan Z, Singh RA, Jayalakshmi S, et al. Location dependence of
microstructure and mechanical properties of Cu–Al alloy fabricated by dual wire CMT.
Mater Res Express 2019;6:126567. [Link]

[15] Dong B, Cai X, Lin S, Fan C. Microstructures and mechanical properties of wire arc
additive manufactured 5183-Al: Influences of deposition dimensions. CIRP J Manuf Sci
Technol 2021;35:744–52. [Link]

[16] Wahsh LM, ElShater AE, Mansour AK, Hamdy FA, Turky MA, Azzam MO, et al.
Parameter selection for wire arc additive manufacturing (WAAM) process. Mater Sci
Technol 2018, MS T 2018 2018:78–85. [Link]

[17] Naveen Srinivas M, Vimal KEK, Manikandan N, Sritharanandh G. Parametric

optimization and multiple regression modelling for fabrication of aluminium alloy thin
plate using wire arc additive manufacturing. Int J Interact Des Manuf 2022.
[Link]
[18] Campatelli G, Campanella D, Barcellona A, Fratini L, Grossi N, Ingarao G.
Microstructural, mechanical and energy demand characterization of alternative WAAM
techniques for Al-alloy parts production. CIRP J Manuf Sci Technol 2020;31:492–9.
[Link]

[19] Wacker C, Köhler M, David M, Aschersleben F, Gabriel F, Hensel J, et al. Geometry and
distortion prediction of multiple layers for wire arc additive manufacturing with artificial
neural networks. Appl Sci 2021;11. [Link]

[20] Saravanakumar S, Sathiyamurthy S, Pathmanaban P, Devi P. Integrating machine learning

and response surface methodology for analyzing anisotropic mechanical properties of
biocomposites. Compos Interfaces 2024;31:1–28.
[Link]

[21] Sengottaiyan S, Subbarayan S, Viswanathan V, Pugazhendi P. Optimized machine

learning with hyperparameter tuning and response surface methodology for predicting
tribological performance in bio-composite materials. Polym Compos 2024:9421–39.
[Link]

[22] Saravanakumar S, Sathiyamurthy S, Vinoth V. Enhancing machining accuracy of banana

fiber-reinforced composites with ensemble machine learning. Meas J Int Meas Confed
2024;235:114912. [Link]

[23] Saravanakumar K, Ananth S, Dharshini M, Medoline J. Investigation of Microstructure

and Mechanical Properties of Aluminium 5356 Using Wire Arc Additive Manufacturing
2023;2:82–98. [Link]

[24] Suganeswaran K, Muthukumar P, Sathiskumar R, Parameshwaran R, Nithyavathy N.

Evaluation of microstructural, tribological and tensile characteristics of AA7075 surface
composites fabricated through friction stir process. Proc Inst Mech Eng Part E J Process
Mech Eng 2023:09544089231170987. [Link]

[25] Prabu G, Jeyaprakash N, Yang CH, Radhika N. Analysis on the transformation of melt
wear to self-healing crack of wire arc additive manufactured Al 5356 alloy. Tribol Int
2023;188:108802. [Link]

[26] Karayel E, Bozkurt Y, Ozdemir C, Kalender M. Wire Arc Additive Manufacturing Of

 5356 Aluminum Alloy. Res Sq 2022:1–16.

[27] Li S, Zhang LJ, Ning J, Wang X, Zhang GF, Zhang JX, et al. Comparative study on the
microstructures and properties of wire+arc additively manufactured 5356 aluminium alloy
with argon and nitrogen as the shielding gas. Addit Manuf 2020;34:101206.
[Link]

[28] Jiangang P, Bo Y, Jinguo G, Yu R, Hongjun C, Liang Z, et al. Influence of arc mode on

the microstructure and mechanical properties of 5356 aluminum alloy fabricated by wire
arc additive manufacturing. J Mater Res Technol 2022;20:1893–907.
[Link]

[29] Kotlarski G, Ormanova M, Nikitin A, Morozova I, Ossenbrink R, Michailov V, et al.

Microstructural and Mechanical Properties of CAP-WAAM Single-Track Al5356
Specimens of Differing Scale. Machines 2024;12:1–15.
[Link]

[30] Wang P, Zhang H, Zhu H, Li Q, Feng M. Wire-arc additive manufacturing of AZ31

magnesium alloy fabricated by cold metal transfer heat source: Processing, microstructure,
and mechanical behavior. J Mater Process Technol 2021;288:116895.
[Link]

[31] Aldalur E, Suárez A, Veiga F. Metal transfer modes for Wire Arc Additive Manufacturing
Al-Mg alloys: Influence of heat input in microstructure and porosity. J Mater Process
Technol 2021;297. [Link]

[32] Wieczorowski M, Pereira A, Carou D, Gapinski B, Ramírez I. Characterization of 5356

Aluminum Walls Produced by Wire Arc Additive Manufacturing (WAAM). Materials
(Basel) 2023;16. [Link]

[33] Zhang Z, Shen J, Hu S, Chen Y, Yin C, Bu X. Optimization of CMT Characteristic

Parameters for Swing Arc Additive Manufacturing of AZ91 Magnesium Alloy Based on
Process Stability Analysis. Materials (Basel) 2023;16.
[Link]

[34] Gierth M, Henckell P, Ali Y, Scholl J, Bergmann JP. Wire Arc Additive Manufacturing
(WAAM) of aluminum alloy AlMg5Mn with energy-reduced Gas Metal Arc Welding
 (GMAW). Materials (Basel) 2020;13:1–22. [Link]

[35] Sahul M, Sahul M, Bočáková B, Kolařík L, Němec T, Kolaříková M. Effect of welding

mode on selected properties of additively manufactured AA5087 aluminium alloy parts. J
Phys Conf Ser 2024;2712. [Link]

[36] Le VT, Mai DS, Doan TK, Paris H. Wire and arc additive manufacturing of 308L stainless
steel components: Optimization of processing parameters and material properties. Eng Sci
Technol an Int J 2021;24:1015–26. [Link]

[37] Tawfik MM, Nemat-Alla MM, Dewidar MM. Effect of Travel Speed on the Properties of
Al-Mg Aluminum Alloy Fabricated by Wire Arc Additive Manufacturing. J Mater Eng
Perform 2021;30:7762–9. [Link]

[38] Li S, Zhang LJ, Ning J, Wang X, Zhang GF, Zhang JX, et al. Microstructures and
mechanical properties of Al-Zn-Mg aluminium alloy samples produced by wire + arc
additive manufacturing. J Mater Res Technol 2020;9:13770–80.
[Link]

[39] Vinoth V, Sathiyamurthy S, Saravanakumar S, Senthilkumar R. Integrating response

surface methodology and machine learning for analyzing the unconventional machining
properties of hybrid fiber-reinforced composites. Polym Compos 2024;45:6077–92.
[Link]

[40] Sathiyamurthy S, Saravanakumar S, Vinoth V. Enhancing tribological performance of

hybrid fiber-reinforced composites through machine learning and response surface
methodology. J Reinf Plast Compos 2024. [Link]

[41] Saravanakumar S, Sathiyamurthy S, Ananthi N, Devi P. Optimization of drilling

characteristics of Al2O3 and boiled eggshell filler-added hybrid bio composite from
agriculture residue. Biomass Convers Biorefinery 2023. [Link]
023-04572-4.

[42] Sathiyamurthy S, Saravanakumar S, Ananthi N, Devi P. Optimization of fiber length and

filler content for improving the mechanical behaviour of musa acuminate fiber-reinforced
epoxy composite using response surface methodology. J Ceram Process Res 2023;24:683–
92. [Link]
[43] Valkov S, Kotlarski G, Ormanova M, Doynov N, Ossenbring R, Petrov P, et al.
Crystallography and mechanical properties of wire arc additively manufactured Al5356
components. Comptes Rendus L’Academie Bulg Des Sci 2021;74:820–7.
[Link]

[44] Fu R, Tang S, Lu J, Cui Y, Li Z, Zhang H, et al. Hot-wire arc additive manufacturing of

aluminum alloy with reduced porosity and high deposition rate. Mater Des
2021;199:109370. [Link]

[45] Evstifeev A, Volosevich D, Smirnov I, Yakupov B, Voropaev A, Vitokhin E, et al.

Comparative Study of the Relationship between Microstructure and Mechanical Properties
of Aluminum Alloy 5056 Fabricated by Additive Manufacturing and Rolling Techniques.
Materials (Basel) 2023;16. [Link]

[46] Li K, Klecka MA, Chen S, Xiong W. Wire-arc additive manufacturing and post-heat
treatment optimization on microstructure and mechanical properties of Grade 91 steel.
Addit Manuf 2021;37:101734. [Link]