Formulation of empirical models for multi-objective optimization of parameters during friction stir welding of AZ80AMg alloy for improving tensile strength and impact toughness

ABSTRACT

An experimental investigation was performed for improving impact toughness and tensile strength of AZ80A Mg alloy plates during FSW, by formulating empirical statistical models via response surface methodology. An inclusive multi–objective optimization strategy incorporating graphical and statistical techniques was also employed for determining important process parameters. A Central Composite Design consisting of 15 experimental runs was employed to systematically evaluate effects of traverse speed, shoulder diameter and rotational speed, on mechanical performance of joints. Experimental inferences revealed that tool’s traverse speed played a dominant role in impacting impact toughness and tensile strength, followed by tool’s shoulder diameter and rotational speed. Interaction amidst traverse and rotational speed had impacted tensile strength, whereas traverse speed and shoulder diameter’s interaction had played a vital role in ascertaining impact toughness. 1.246 mm/sec traverse speed, 12.868 mm shoulder diameter, 1021 rpm was identified as optimized parameters and flaw free joints fabricated at this combination exhibited 170.48 MPa tensile strength and 23.41 J impact toughness. Perfect accuracy of the formulated statistical model was confirmed by validation experimental results, exhibiting negligible discrepancy amidst anticipated and actual runs. SEM analysis of the fractured specimen of the flaw free joints exhibited a ductile failure, demonstrating superior plastic deformation capability.

Keywords:
AZ80A Mg alloy; Friction stir welding; ANOVA; Response surface methodology; Tensile Strength; Impact toughness

1. INTRODUCTION

In the recent years, regulations framed by governments of several nations and environmental concerns for designing vehicles with minimized consumption of fuel and energy, had put forward a confrontation to the automobile industries to manufacture light in weight automotive, aerospace structures and parts. Due to this scenario, the demand for light in weight metals including alloys of Mg (i.e., magnesium) had gained wider attraction for replacing the traditional materials used in the fabrication of automotive parts [¹, ²]. It was recorded that alloys of Mg have replaced plastic materials in computer & electronic industries, steel and aluminium in aerospace and automotive sectors, owing to their attractive properties including superior strength to weight proportion, perfect damping capability contributing towards minimized vibration and noise, desirable density, enhanced recyclability, ease degree of manufacturability, effecting shielding against electro-magnetic interferences, ability to get casted easily etc., [³–⁵]. On the contrary, the usage of Mg alloys in above mentioned industries have been limited owing to their inferior ductility & formability, arising due to their HCP (i.e., hexagonally closed packed) architectures, poor hardness, insufficient resistivity against wear and lower1 strength [⁶–⁸].

Structures, parts and components fabricated out of welding process are inevitable in automotive and aerospace industries. At the same time, alloys of Mg exhibit a poor degree of weldability, owing to their extreme electrical, thermal attributes and oxidation characteristic features. When Mg alloys are subjected to welding using conventional welding techniques, the fabricated weldments were found to be prone to splash, slags, pores, residual stresses and several other flaws, thereby deteriorating the strength of the fabricated component [⁹]. This limitation in welding of alloys of Mg hinders the wide usage of Mg alloys in the automotive and aerospace industries. So, there prevails an inevitable need for identifying a suitable welding process for joining together alloys of Mg and to examine the joining behavior of the welded Mg alloys [¹⁰, ¹¹].

FSW (i.e., friction stir welding) is one of the unique, environmental friendly solid state welding process was found to be effective alternative to traditional welding methodologies and as the metal subjected to FSW process gets welded, before reaching its melting temperature, it can be employed for joining together alloys of Mg. Moreover, during the FSW process, there is no need for a shielding medium, wire or arc, it does not generate any fumes and oxidation does not occur. It was also reported by experimental researchers that the joints fabricated using the FSW process are free from several defects including blow holes, splash, spatter, porosity, cracks etc [¹²–¹⁴]. In addition to these, employing FSW process for joining together Mg alloys will also lead to minimization of residual stresses, owing to the uniform frictional heat and solid state category of bonding. Excessive loss of energy is also prevented during the FSW process, as the perfect control of input of frictional heat contributes for sufficient softening of plasticized material. Another unique feature of FSW is that the entire process can be automated in an easier manner, thereby enabling larger volume fabrication of automotive and aerospace components, parts [¹⁵–¹⁷].

At the same time, there prevails several parameters which play a key role in ascertaining the properties and quality of the joints attained using the FSW process. For instance, the parameters including speed of rotation of the tool, its speed of traverse, force being employed axially on the tool, angle at which tool is tilted, geometry of the tool pin, dimeter of the tool shoulder etc., impact the flow of the softened material, volume of frictional heat generated, consolidation & mixing of the plasticized metal, distribution & conduction of frictional heat, flow behavior of the softened metal etc., and as a result play a major role in determining the micro structural characteristics and mechanical, thermal properties of the fabricated joints [¹⁸–²⁰]. As a result, there prevails an inevitable need for employing these parameters in an ideal combination, so as to attain superior quality, flaw free joints with appreciable mechanical attributes, uniformity in micro structure.

On the other hand, it is a crucial task to optimize these parameters of the FSW process, especially during the FSW of superior materials including alloys of Mg (like AZ80A), titanium and during FSW of entirely distinctive metals. FSW of these superior metals without optimizing the relevant parameters were found to be leading to the generation of inferior quality joints inheriting several flaws including insufficient fusion, excessive thinning, surface galling, expulsion of plasticized metal, formation of voids, worm holes etc., [²¹,²²,²³]. Various conventional approaches being used for identifying the ideal combination of parameters of FSW process basically depend on experimental strategies, where the parameters are modified iteratively based on the recorded joint quality [²⁴]. Compared to aerospace-grade 2050 Al–Li alloy, AZ80A Mg alloy exhibits lower melting temperature, higher thermal sensitivity, and a narrower plastic deformation window during friction stir welding. These characteristics increase its susceptibility to heat-induced defects and microstructural instability, thereby necessitating systematic multi-objective optimization of process parameters [²⁵].

For instance, the traditional optimization technique based on trial and error process was proven to consume lot of time, require intensive resources and relies largely on the expertise of the operator, leading to the generation of variable, uncertain results. Analysis of single factor optimization approach isolates the impact of each & every parameter, by keeping all other parameters as constant values [²⁶]. Even though, this conventional methodology provided fruitful insights about suitable identification of parameters, this methodology usually demands for large quantum of experimental results data and high expertise for interpreting these results in an effective manner. Although, the trial investigational runs are minimized by employing these conventional numerical optimization methodologies, these techniques usually fail to ascertain the non-linear interactive relationship prevailing among multiple process parameters, resulting in substandard parameter combinations [²⁷–²⁹]. Moreover, these traditional methodologies were found to be unsuitable for FSW of superior materials like Mg alloys (like AZ80A) possessing complicated material flow and thermal attributes. Several drawbacks of these traditional techniques including its incapability to adopt to automated process like FSW have restricted their adaptability and scalability in several sectors including automotive and aerospace industries, thereby generating an inevitable demand for identifying improved, effective optimization techniques for overcoming the above mentioned drawbacks in an efficient manner [³⁰–³²].

Among the several available optimization methodologies, DoE (i.e., design of experiments) based RSM (i.e., response surface methodology) will be perfectly suitable for optimizing the parameters during FSW of AZ80A Mg alloy plates [³³, ³⁴]. For instance, DoE permits a well-organized architecture for performing experimental runs involving minimized resources and maximized output data. Investigating the individual impact of several process parameters and their interactive impacts is also possible with DoE, which is very much essential during FSW of AZ80A Mg alloy plates where the interdependencies amidst the process parameters reasonably impact the quality of the joints [³⁵]. Moreover, the attained experimental results can also be evaluated in a statistical manner while employing DoE for optimization, thereby assuring reproducibility and reliability, which will be very much advantageous during FSW of AZ80A Mg alloy plates, where the variation in the material properties will complicate the optimization process [³⁶, ³⁷]. The systematic strategy of DoE permits the researchers to explore the parameter’s impact in a comprehensive manner, thereby helps them to identify the significant parameters (which affect the joint quality) in an effective manner.

On the other hand, using the foundation laid by DoE, using RSM numerical models portraying the relationship amidst the outcomes and input parameters, can be formulated. By establishing quadratic equations, RSM can provide a comprehensive view of FSW dynamics (especially non-linear and thermal behavior exhibited by Mg alloy during FSW), by capturing both the non-linear as well as linear impacts, relationships amidst the parameters [³⁸, ³⁹]. RSM’s capability to anticipate the optimized combination of FSW process parameters with perfect accuracy will lower down the experimental run costs and by observing the response surface graphs, investigators can visualize the impact of parameters on the various properties including density of defect, impact toughness, tensile strength etc., [⁴⁰, ⁴¹].

In this research paper, it had been planned to combine DoE with RSM, as this combination of optimization architecture will be very much effective, combining the merits of both the strategies. Effective collection of input data and analyzing the interaction amidst the parameters will be ensured by DoE and RSM will identify the ideal parameters combination by generating perfect forecasting models. This synchronized approach will be very much essential for handling the various technical constraints related with FSW of AZ80A Mg alloy plates, owing to the metal’s superior susceptibility and reactivity to thermal deformations. In this work, the FSW parameters namely tool’s traverse speed, its speed of rotation and its shoulder diameter are optimized with the objective of enhancing the impact toughness and tensile strength of the fabricated joints.

2. EXPERIMENTAL FRAMEWORK

2.1. Base metal

In this work, rectangular plates (105 × 55 × 6 mm) of AZ80A Mg alloy was taken as the base metal of research. This Mg alloy contains zirconium and zinc as its major alloying constituents. Table 1 describes in detail, the several chemical ingredients and mechanical properties of this Mg alloy.

Table 1
Constituents and properties of the base metal.

PARENT METAL	Zn	Al	Fe	Ni	Cu	Mn	Si	Mg	Elongation %	Tensile Strength, MPa	Impact Toughness J
AZ80A Mg alloy	0.65	8.1	0.005	0.049	0.05	0.12	0.1	Balance	3.5	205	30

Figure 1(a) portray the optical micro-structural image of the base metal, namely AZ80A Mg alloy revealing the presence of non-uniformly scattered, irregular grain architectures and this can also be observed very clearly in the SEM (i.e., scanning electron microscope) image of this base metal, as seen in the Figure 1(b), which illustrates an unevenly dispersed grain architecture within the matrix of the Mg alloy.

Figure 1
Base metal’s (a) Micro-structural Image and (b) SEM Image.

The base metal’s micro-structure can be described as an unsteady structure of Mg based solid solutions and granular MgZn₂ intermetallic phase constituents scattered along the boundaries of the grain. Inconsistent and coarse distribution of these needle like MgZn₂ structural deformities not only reduces the alloy’s resistivity against fatigue & ductility, but also offers reasonable challenge in attaining flaw free, superior quality joints [⁴²]. Its unique capability to refine irregular grain architecture, abolish granular precipitates, rearrange alloying ingredients, FSW process will provide a perfect solution to the above mentioned issue during the joining of flat plates of AZ80A Mg alloy [¹⁸, ²³]. At the same time, the intricacy in FSW process arising due to interdependencies amidst parameters demands for employing an enhanced optimization technique for determining the ideal process parameters.

2.2. Machine and tool

In this work, a semi-automatic type FSW machine as seen in the Fig. 2(a) was used for welding together the rectangular plates of AZ80A Mg alloy as a butt joint configuration, by means of employing cylindrically tapered pin profiled tools (possessing three different shoulder diameters) fabricated out of M42 grade high speed steel [⁴³]. In this research work, the rectangular plates of AZ80A Mg alloy were held together firmly by placing them in a specially designed jig and fixture assembly as seen in the Figure 2(b) & (c). Photos of cylindrically tapered pin profiled tools (possessing three different shoulder diameters) can be seen in the Figure 2(d)–(f).

Figure 2
(a) Semi-automatic FSW Machine, (b) & (c) specially designed jig and fixture assembly and (d), (e) & (f) cylindrically tapered pin profiled tools possessing three different shoulder diameters used in this work.

2.3. Trial experimental runs

In this work, initially trial experimental runs were conducted to ascertain the appropriate working range of the parameters of the FSW process. Three distinctive parameters namely, rotational and traverse speed of the tool, its shoulder diameter were taken into consideration and several trial experimental runs were carried out by employing distinctive combinations of these parameters. During these trial runs, rectangular plates of AZ80A Mg alloy were friction stir welded by altering one of these parameters from their maximum to minimum value, with other parameters being kept constant. The operational spectrum of each & every chosen parameter was determined by taking into account, several factors like formation of flash, flaws visible at macro level like surface grooves, shoulder mark flaws, worm holes etc. Table 2 describes in brief about the important observations made by examining the macro-structure of the AZ80A Mg alloy joints fabricated during these trial experimental runs.

Table 2
Macro-structural observations of the joints attained during trial runs.

PROCESS PARAMETERS	WORKING RANGE	MACRO-STRUCTURAL OBSERVATION	FLAW DESCRIPTION	PROBABLE REASON
Shoulder diameter	< 11 mm		Tunnel flaws	Minimized flow of metal had generated tunnels at the bottom of the nugget zone
	> 16 mm		Excessive Flash	Excessive heat and metal flow due to employment of larger tool diameter had extruded the excess metal from the joint region.
Tool’s speed of rotation	< 890 rpm		Insufficient material mixing	Inadequately generated heat had led to improper mixing of metals in the weld zone, and had resulted in poor bonding
	> 1610 rpm		Voids	Fast stirring action due to larger rotational speeds had trapped air and had generated voids in the nugget zone
Tool’s speed of traverse	< 0.6 mm/sec		Kissing bond	Lower traverse speed generated insufficient volume of heat, failing to disrupt oxide layers and had led to weak metallurgical bonding
	> 1.4 mm/sec		Surface lack of fill	Faster tool traverse speed had reduced the time of contact amidst the metal &

2.4. Parameters and their levels

Table 3 describes the selected levels of the significant parameters of the FSW process taken into consideration in this work. Notations and units of these parameters are described in the Table 3.

Table 3
Significant parameters, their units and chosen levels.

PARAMETER OF FSW PROCESS	NOTATION	UNIT	LEVELS
			–1.414	–1	0	+1	+1.414
Traverse Speed	T	mm/sec	0.65	0.75	1	1.25	1.35
Rotational Speed	R	rpm	896	1000	1250	1500	1603
Shoulder Diameter	S	mm	11.4	12	13.5	15	15.6

The above mentioned feasible limits of the parameters of the FSW process were chosen with the objective of fabricating flaw free AZ80A Mg alloy joints. In this research, the following process parameters were maintained constant throughout all experimental trials to ensure controlled investigation of the selected variables (rotational speed, traverse speed, and shoulder diameter):

• Pin Diameter: 4.8 mm

• Pin Length: 5.75 mm (slightly less than plate thickness to avoid root defects)

• Tool Tilt Angle: 0° (constant)

• Plunge Depth: 0.1 mm (controlled to ensure proper shoulder contact)

• Axial Force: 4 kN

• Backing Plate: Hardened steel backing plate with high thermal conductivity and rigid clamping support

• Welding Side Orientation: Advancing side maintained consistently on the same plate orientation for all trials

These parameters were deliberately kept constant to avoid multivariable interference and to ensure statistical robustness of the DoE–RSM model. The empirical models developed in this work are therefore valid within the defined experimental boundary conditions. As the range of these selected significant process parameters are wide, a central composite (2^nd sequence) rotatable type design, i.e., CCD was employed to formulate the design matrix, because of its capability to efficiently estimate second-order polynomial models and evaluate interaction effects among process parameters. The experimental runs were conducted based on this design matrix. Each parameter was investigated at five coded levels (−1.414, −1, 0, +1, +1.414) to ensure adequate coverage of the design space and to maintain rotatability of the CCD. The devised design matrix constituted a total of 15 codified schematic sets and involved an entire duplication pattern comprising 1-central, 8-factorial, 6 locus points. As the design matrix constituted 5 positions, the larger and smaller boundaries were coded as +1.414 and –1.414 respectively and the existing 3 positions were ascertained as –1, 0 and +1, based on the below mention equation:

(1) $${\mathrm{X}}_{\text{i}}=2[2\text{X}-({\mathrm{X}}_{\mathrm{m}\text{ax}}+{\mathrm{X}}_{\min })]/[{\mathrm{X}}_{\max }-{\mathrm{X}}_{\min }]$$

where fixed coded value for the X variable is denoted by Xi and the range of this variable falls within X_max to X_min. A total of 15 AZ80A Mg alloy joints were fabricated by means of employing FSW process as suggested by the design matrix. 3 test specimen were extracted from each joint fabricated during this work and were subjected to examinations. Table 4 describes the mean of the examination results of these 3 test specimen extracted from AZ80A joints.

Table 4
Outcomes of experimental runs.

S.NO.	CODED VALUES			REALISTIC VALUES			OUTCOMES
	T	R	S	T	R	S	ULTIMATE TENSILE STRENGTH (MPa)	IMPACT TOUGHNESS (J)
1	0	0	0	1	1250	13.5	166	19.2
2	–1	1	1	0.75	1500	15	139.2	16.3
3	–1	–1	–1	0.75	1000	12	133.5	18
4	–1.414	0	0	0.65	1250	13.5	140.7	16.5
5	1	1	–1	1.25	1500	12	153.5	23
6	0	0	1.414	1	1250	15.6	142.5	19.1
7	0	0	0	1	1250	13.5	169.3	19.8
8	0	0	0	1	1250	13.5	162	19.8
9	0	0	–1.414	1	1250	11.4	156.2	22
10	0	–1.414	0	1	896	13.5	158.8	21.2
11	0	0	0	1	1250	13.5	169	19
12	0	1.414	0	1	1603	13.5	150.2	18.4
13	1	–1	1	1.25	1000	15	156.4	21.9
14	0	0	0	1	1250	13.5	171	19.1
15	1.414	0	0	1.35	1250	13.5	162	24

3. ESTABLISHMENT OF EXPERIENTIAL RELATIONS

3.1. Statistical model formulation

With the objective of establishing a perfection relation amidst the outcomes and FSW parameters, in this work, response surface methodology (i.e., RSM) was employed and a prophetical numerical model was developed. Results attained from lack of fit and subsequent sum of squares tests proved that the quadratic regression model is the suitable approach for anticipating the outcomes of this work, namely the ultimate tensile strength (i.e., UTS) and impact toughness (IT). UTS and IT of the fabricated AZ80A Mg alloy joints were anticipated by means of the formulated 2^nd order quadratic relations, making use of the UTS and IT values described in the Table 4. A 2^nd order polynomial equation was employed to describe “Y”, namely the response surface and it is mentioned below:

(2) $$\mathrm{Y}={\mathrm{\beta }}_{\text{0}}+{\displaystyle \sum {\mathrm{\beta }}_{\text{i}}}{\mathrm{X}}_{\text{i}}+{\displaystyle \sum {\mathrm{\beta }}_{\text{i}}{}_{\mathrm{i}}}{\mathrm{X}}_{\text{i}}^2+{\displaystyle \sum {\mathrm{\beta }}_{\text{i}}{}_{\mathrm{j}}}{\mathrm{X}}_{\text{i}}{\mathrm{X}}_{\text{j}}$$

The second-order polynomial model employed in this work enables effective representation of the inherently non-linear thermo-mechanical behavior occurring during friction stir welding of AZ80A Mg alloy. Due to the alloy’s low melting temperature and narrow thermal processing window, tensile strength and impact toughness exhibit peak-type responses with increasing heat input. The quadratic terms (X_i²) in the RSM model capture this curvature behavior, while interaction terms (X_i X_j) represent coupled thermal effects between rotational speed, traverse speed, and shoulder diameter. This quadratic modeling capability allows accurate prediction of thermal saturation, grain coarsening thresholds, and optimum parameter combinations. For three based factors, the chosen regression (i.e., polynomial) was described as

(3) $$\begin{array}{l}\mathrm{U}\text{l}\mathrm{t}\text{i}\mathrm{m}\text{a}\mathrm{t}\text{e}\mathrm{}\text{t}\mathrm{e}\text{n}\mathrm{s}\text{i}\mathrm{l}\text{e}\mathrm{}\text{s}\mathrm{t}\text{r}\mathrm{e}\text{n}\mathrm{g}\text{t}\mathrm{h}(\mathrm{U}\text{T}\mathrm{S})\mathrm{}\text{o}\mathrm{r}\mathrm{I}\text{m}\mathrm{p}\text{a}\mathrm{c}\text{t}\mathrm{}\text{T}\mathrm{o}\text{u}\mathrm{g}\text{h}\mathrm{n}\text{e}\mathrm{s}\text{s}\mathrm{}(\mathrm{I}\text{T})\mathrm{}=\mathrm{}{\mathrm{b}}_{\text{0}}+{\mathrm{b}}_{\text{1}}(\mathrm{T})+{\mathrm{b}}_{\text{2}}(\mathrm{R})+\\ {\mathrm{b}}_{\text{3}}(\mathrm{s})+{\mathrm{b}}_{1\text{2}}\mathrm{T}\text{R}+{\mathrm{b}}_{1\text{3}}\mathrm{T}\text{S}+{\mathrm{b}}_{2\text{3}}\mathrm{R}\text{S}+{\mathrm{b}}_{1\text{1}}{\mathrm{T}}^{\text{2}}+{\mathrm{b}}_{2\text{2}}{\mathrm{R}}^{\text{2}}+{\mathrm{b}}_{33}{\mathrm{S}}^{\text{2}}\end{array}$$

Central composite (2^nd sequence) rotatable type design was employed to attain all the relevant coefficients and this was attained by using Design Expert analytical software and after attaining the important coefficients (possessing a confidence level of 95%), interconnections were established by making use of these coefficients. The Polynomial equations deduced from the experimental run data (described in the Table 4) is given below and these equations describe the final statistical interconnections for estimating the ultimate tensile strength (i.e., UTS) and impact toughness (IT):

(4) $$\begin{array}{l}\mathrm{U}\text{T}\mathrm{S}=1\text{6}7.1\text{9}6+7.5\text{3}0\text{6}9\text{T}-3.0\text{4}0\text{5}6\text{R}-4.8\text{4}3\text{6}8\text{S}+6.9\text{9}3\text{6}8\text{T}\mathrm{R}+3.7\text{4}0\text{5}6\text{T}\mathrm{S}-\\ 1.7\text{6}9\text{3}1\text{R}\mathrm{S}-7.5\text{9}3\text{5}2{\mathrm{T}}^{\text{2}}-6.0\text{1}8\text{5}2{\mathrm{R}}^{\text{2}}+8.5\text{9}3\text{5}2{\mathrm{S}}^{\text{2}}\end{array}$$

(5) $$\begin{array}{l}IT=1\text{9}.5+2.6\text{5}1\text{6}5T-0.9\text{8}9\text{9}4R-1.0\text{2}5\text{3}S-0.3\text{2}5\text{3}0\text{5}TR-0.8\text{3}9\text{9}4\text{9}TS+\\ 0.0\text{0}1\text{6}5\text{0}4\text{3}RS+0.2\text{2}5T2-0.0\text{0}1\text{7}3\text{6}1R^2+0.3\text{7}5S^2\end{array}$$

3.2. Formulated model – Competency verification

The sufficiency of the formulated experimental interconnections was evaluated by means of employing ANOVA (i.e., analysis of variance). The adequacy of the formualted 2^nd order regression models for ultimate tensile strength (UTS) and impact toughness (IT) was evaluated using ANOVA, is presented in Tables 5 & 6.

Table 5
ANOVA results for UTS.

SOURCE	SUM OF SQUARES	DOF	MEAN SQUARE	VALUE OF F	VALUE OF P	CONTRIBUTION
Developed Model	1934.45	9	214.94	19.57	0.0022	Significant
T-Traverse speed	226.84	1	226.84	20.65	0.0061	11.40%
R-Rotational speed	36.98	1	36.98	3.37	0.1260	1.86%
S-Shoulder diameter	93.85	1	93.85	8.54	0.0329	4.72%
TR	97.82	1	97.82	8.91	0.0307	4.92%
TS	27.98	1	27.98	2.55	0.1714	1.41%
RS	6.26	1	6.26	0.5699	0.4843	0.31%
T2	444.82	1	444.82	40.49	0.0014	22.36%
R2	279.43	1	279.43	25.44	0.0040	14.05%
S2	569.69	1	569.69	51.86	0.0008	28.64%
Residual	54.93	5	10.99
Lack of Fit	4.69	1	4.69	0.3738	0.5740	Not significant
Pure Error	50.23	4	12.56
Cor Total	1989.38	14
Standard deviation	3.31		R2	0.9724
Mean	155.35		Adjusted R2	0.9227
C.V.%	2.13		Anticipated R2	0.7768
Press	583.13		Adeq Precision	12.6952

Table 6
ANOVA results for IT.

SOURCE	SUM OF SQUARES	DOF	MEAN SQUARE	VALUE OF F	VALUE OF P	CONTRIBUTION
Developed Model	67.82	9	7.54	23.85	0.0014	Significant
T-Traverse speed	27.83	1	27.830	88.07	0.0002	40.10%
R-Rotational speed	3.92	1	3.920	12.41	0.0169	5.65%
S-Shoulder diameter	4.2	1	4.200	13.29	0.0148	6.05%
TR	0.2116	1	0.212	0.67	0.4504	0.30%
TS	1.41	1	1.410	4.46	0.0884	2.03%
RS	0.1000	1	0.100	0.32	0.5980	0.14%
T2	0.3905	1	0.391	1.24	0.3169	0.56%
R2	0.2004	1	0.200	0.63	0.4619	0.29%
S2	1.08	1	1.080	3.42	0.1238	1.56%
Residual	1.58	5	0.316
Lack of Fit	0.972	1	0.972	3.08	0.1398	Not significant
Pure Error	0.608	4	0.152
Cor Total	69.40	14
Standard deviation	0.5621		R2	0.9772
Mean	19.82		Adjusted R2	0.9363
C.V.%	2.84		Anticipated R2	0.8125
Press	106.02		Adeq Precision	16.34

For tensile strength, the model is statistically significant (p = 0.0022) with a coefficient of determination R² = 0.9724, indicating that 97.24% of the variability in tensile strength is explained by the selected process parameters within the investigated design domain. The adjusted R² (0.9227) confirms that the model maintains strong explanatory capability after accounting for the number of predictors. The predicted R² (0.7768) shows reasonable agreement with the adjusted R², suggesting acceptable predictive performance without excessive overfitting [²⁹]. For impact toughness, the developed model is also statistically significant (p = 0.0014), with R² = 0.9772 and adjusted R² = 0.9363, demonstrating high explanatory strength. The predicted R² (0.8125) is in satisfactory agreement with the adjusted value, confirming model reliability for interpolation within the design space [⁴⁴].

Based on the chosen significance level (α = 0.05), traverse speed (T) is the dominant statistically significant factor for both UTS and IT. For UTS, shoulder diameter (S), the TR interaction, and quadratic terms (T², R², S²) are also significant contributors (p < 0.05), whereas rotational speed (R) and certain interaction terms (TS and RS) are statistically insignificant within the studied parameter range [²⁸]. For IT, traverse speed, rotational speed, and shoulder diameter are statistically significant, while interaction and quadratic terms exhibit p-values greater than 0.05 and are therefore interpreted cautiously. The lack-of-fit for both models is statistically insignificant (p > 0.05), indicating that the residual error is primarily due to experimental variability rather than model inadequacy. Adequate precision values greater than 4 further confirm a satisfactory signal-to-noise ratio. Thus, the statistical conclusions presented are strictly based on significant ANOVA terms, ensuring analytical rigor and avoiding over-interpretation of non-significant effects [²⁶, ³³]. With the objective of evaluating whether the residuals follow a normal distribution, normal probability residual plots have been generated for all responses (both for UTS & IT). The plots illustrated in the Figure 3(a) & (b) demonstrate that the residual points lie approximately along a straight line, indicating that the residuals are normally distributed and that the model assumptions are satisfied.

Figure 3
Normal probability residual plots for the responses: (a) UTS and (b) IT.

Furthermore, plots comparing predicted versus experimental values have been added for each response variable, namely UTS and IT. These plots illustrated in the Figure 4(a) & (b) show a strong agreement between the predicted values obtained from the RSM models and the corresponding experimental results. It can also be observed that the data points are closely distributed around the 45° reference line, confirming that the developed models accurately represent the experimental behavior within the studied design space.

Figure 4
Graphical illustration of actual vs. predicted values of (a) UTS and (b) IT.

Moreover, ANOVA results presented in Tables 5 and 6 indicate that the developed quadratic models for both ultimate tensile strength (UTS) and impact toughness (IT) are statistically significant at the selected significance level (α = 0.05), as evidenced by model p-values below 0.05. For tensile strength, traverse speed and shoulder diameter, along with the TR interaction and quadratic terms, are statistically significant contributors (p < 0.05), whereas rotational speed and certain interaction terms are not significant within the investigated range. For IT, traverse speed is the dominant significant factor, followed by rotational speed and shoulder diameter, while interaction and quadratic terms show p-values greater than 0.05 and are therefore not interpreted as statistically influential. The lack-of-fit for both models is statistically insignificant (p > 0.05), confirming model adequacy.

4. EXPERIMENTAL INFERENCES AND DISCUSSIONS

By examining the micro-structural images of various zones of AZ80A joints it can be understood that employment of appropriate combination of FSW parameters had altered the base metal’s architecture and this had occurred owing to the combined impact of the severe plastic deformation and peak temperature arises during FSW. It was recorded by several investigators that the morphological modification in the nugget zone of the friction stir welded joints as seen in the Figure 5 was due to the severe dynamic recrystallization and plastic deformation [¹⁹, ⁴⁵].

Figure 5
Nugget zone of the friction stir welded joint exhibiting the presence of an onion ring structure.

This modified nugget zone usually found to resemble an onion ring structure as seen in the Figure 3 and this onion ring structure occurs mainly because of the cyclically deposited deformed metal layers as the tool slides in forward direction. This morphology modification was mainly due to the frictional heat generated and the peak temperature being attained during the FSW process [²⁴, ³⁴]. This frictional heat and peak temperature was proven to be influenced to a major extent by the tool’s traverse, rotational speed and its shoulder diameter. The forthcoming sections discusses in detail the impact of these process parameters on the impact toughness and tensile strength of the friction stir welded AZ80A Mg alloy joints.

4.1. Impact of traverse speed

The perturbation graphs being attained for impact toughness and tensile strength of the friction stir welded AZ80A Mg alloy joints are illustrated in the Figure 6(a) & (b) respectively. These graphs help us to visualize the impacts of the various process parameters of FSW, at the center region of the design span.

Figure 6
Perturbation graphs for (a) impact toughness and (b) tensile strength of the friction stir welded AZ80A Mg alloy joints.

From the experimental run outcomes, it can be understood that the employment of lower values of tool traverse speed have resulted in the fabrication of AZ80A Mg alloy joints possessing inferior values of impact toughness and tensile strength. This was due to the long exposure to the generated frictional heat and deformation impacts. Exposure of MgZn₂ intermetallic precipitates to enormous volume of frictional heat for a longer time, had coarsened these intermetallic precipitates, thereby reducing the effectiveness of the strengthening of the precipitates, which indirectly deteriorated the tensile strength of the joints [¹³, ⁴⁶]. In addition to this, the exposure to excessive heat for longer period, have led to the inordinate dissolution of MgZn₂ precipitates, thereby reducing the strengthening of precipitates, followed by generation of micro-voids. These micro-voids had served as sites of crack initiation and have reduced the impact toughness of the joints [⁴, ⁴⁷].

Inadequate input of frictional heat due to employment of larger traverse speeds have led to improper softening of materials and inferior flow of those materials. As a result, the MgZn₂ precipitates in the joints fabricated at higher traverse speeds have dissolved insufficiently, had weakened the boundaries of the grain. This improper hardening of the precipitates has reduced the Mg alloy joint’s resistivity against deformation. Due to the poor bonding at the interface region, kissing bonds have been observed in the joints, which have deteriorated the tensile strength of these joints. at the same time, the energy absorption capability of the joints during impact was reduced due to the presence of coarse grains and had reduced the resistivity of the fabricated joints against fracture [²⁶, ³⁴]. By examining the joints fabricated during the experimental runs it can be inferred that the joint fabricated during tool traverse speeds of 1 mm/sec and 1.25 mm/sec have exhibited larger values of tensile strength. At the same time, with the increase in the welding speed, the fabricated joints impact toughness has also increased.

4.2. Impact of rotational speed

Impact toughness is a significant property that ascertain the AZ80A Mg alloy’s capability to imbibe energy before encountering fracture under abrupt loading scenarios. In our research work, it can be inferred from the Figure 6(a) that the impact toughness of the friction stir welded AZ80A Mg alloy joints had declined with the escalation in the rotational speed. This was mainly due to the fact that during larger rotational speeds, the frictional heat is generated in excess volumes and this excessive heat had lowered the degree of dynamic recrystallization, leading to the formation of coarse, elongated grain structures. Specifically, grain coarsening at higher rotational speeds reduces the total grain boundary area per unit volume, thereby decreasing grain boundary density. Since grain boundaries act as effective barriers to crack initiation and propagation as well as dislocation motion (Hall–Petch relationship), a reduction in grain boundary density diminishes resistance to impact loading [²¹]. SEM micrographs presented in Figure 7(a) clearly show coarser and elongated grain morphologies at higher rotational speeds, whereas joints fabricated at optimal rotational speeds (1200–1250 rpm) exhibit finer, equiaxed dynamically recrystallized grains. The refined microstructure increases grain boundary area and enhances crack deflection capability, thereby improving energy absorption during impact [²⁴, ³²].

Figure 7
SEM images illustrating the impact of (a) lower rotational speed and (b) larger rotational speed on the grain structures.

Generation of frictional heat, flow of the plasticized metal and refinement of grain structures during the FSW process was highly impacted by the tool’s rotational speed. By examining the friction stir welded AZ80A Mg alloy joints, it can be inferred that the tensile strength escalated with the escalation in the tool’s rotational speed from 896 rpm to 1250 rpm and started to decline with the further escalation in the rotational speed, as illustrated in the Figure 6(b). Employment of larger rotational speeds causes generation of frictional heat in an excessive manner and this heat generated in an excessive manner leads to improper growth of the grains, softens the heat impacted zone and as a result, the tensile strength of the fabricated joints had reduced [¹⁷, ⁴⁸]. This scenario can be visualized in the SEM image displayed in the Figure 7(b). Another reason for decline in the tensile strength of the joints fabricated at larger rotational speeds is that the large volume of frictional heat generated degrades the gradient of hardness of the fabricated joints, which in turn weakens the nugget zone [²⁹].

4.3. Hall–petch validation

The SEM micrographs illustrated in the Figure 7(a) & (b) clearly demonstrate that rotational speed critically governs grain evolution within the nugget zone (NZ) of friction stir welded AZ80A Mg alloy joints. Under lower rotational speed, dynamic recrystallization produced uniformly distributed equiaxed grains with an average size of 3.6 ± 0.8 μm, whereas higher rotational speed resulted in significant grain coarsening with an average size of 9.8 ± 2.4 μm. This nearly 2.7-fold increase in grain size indicates excessive heat input and prolonged thermal exposure at elevated rotational speed. The mechanical implications of this refinement were quantitatively evaluated using the Hall–Petch relationship:

(6) $${\sigma }_y={\sigma }_0+k{d}^{-\raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$\text{2}$}\right.}$$

Assuming a representative Hall–Petch slope for Mg alloys (k ≈ 200 MPa·μm^1/2), the calculated strengthening increment between the refined (3.6 μm) and coarse (9.8 μm) conditions is approximately 40 MPa. Although this theoretical value strictly applies to yield strength, it confirms that grain refinement is a dominant contributor to the experimentally observed improvement in tensile performance under optimized rotational speed. The refined microstructure increases grain boundary density, thereby enhancing resistance to dislocation motion and delaying localized plastic instability [²¹, ⁴⁹]. The transverse hardness profile across various zones further substantiates this correlation. The optimized nugget zone exhibited peak hardness values (~80–82 HV), significantly higher than the base metal (~65 HV). Given the proportionality between hardness and yield strength (HV ≈ 3σ_y), this increase aligns with Hall–Petch strengthening. Conversely, the heat impacted zone displayed softening (~58–60 HV), attributable to grain growth and thermal overexposure. Impact toughness trends also correlate strongly with grain size [¹⁵]. The refined grains at lower rotational speed promote crack deflection and energy dissipation through increased grain boundary area, enhancing resistance to rapid crack propagation. In contrast, coarse grains formed at higher rotational speed reduce crack path tortuosity and diminish impact energy absorption. In short, the integrated grain size quantification, hardness mapping, and Hall–Petch validation establishes a clear and statistically supported structure–property relationship, demonstrating that optimized rotational speed maximizes dynamic recrystallization while suppressing detrimental grain growth, thereby improving tensile strength and impact toughness simultaneously.

4.4. Impact of diameter of tool shoulder

Two major mechanisms responsible for flow of the frictional heat during the FSW process are the flow of heat from the tool’s shoulder and the flow of heat from the tool’s pin. It was proven by researchers that the flow of frictional heat from the pin is reasonably small when compared with that from the shoulder. The movement of the plasticized metal around the upper portion of the nugget zone (i.e., roughly the top 1/3^rd) is impacted to a greater extent by the tool’s shoulder rather than the profile of the tool pin. The shoulder diameter can be given more importance than the tool pin owing to the facts that the area of the shoulder surface is very larger when compared with that of the area of the pin surface, the diameter of the tool shoulder is highly proportional to the volume of frictional heat being generated, the torque & behavior of plasticized metal is dependent on the peak temperature being generated [²⁹, ⁵⁰]. As a result, in this research work, the shoulder diameter is taken into account (neglecting pin diameter) and the torque being generated is contributed to a reasonable extent by the tool shoulder diameter.

From the experimental run outcomes, it was inferred that the peak temperature escalates with the escalation in the diameter of the tool shoulder, with the rotational and traverse speed employed at constant values. Moreover, the stress related plasticized metal flow in the weld zone declines with this shoulder diameter escalation [³⁴]. But at the same time, when the diameter of the tool shoulder is raised above 15 mm, the resistivity exerted by the plasticized metal against its flow around the tool also gets declined and as a result, during the employment of these shoulder diameter values, the tool loses its capability to impact the movement of the plasticized metal.

When tool possessing larger shoulder diameters (greater than 13.5 mm) were employed, larger volumes of frictional heat gets generated owing to the increased area of contact amidst the tool and the surface of the plates to be welded. This in turn widens the heat impacted and thermo-mechanically impacted zone, thereby deteriorating the tensile strength of the AZ80A Mg alloy joints. During employment of tool possessing smaller shoulder diameters (11–12 mm), frictional heat was found to be generated in minimal volumes, due to which complete recrystallization of grains could not happen. This incomplete mixing of partially recrystallized grains had weakened the interfacial strength of the bonding structures. As a result of this weak bonding and improperly recrystallized metals, the joint exhibited a poor tensile strength [¹⁵, ⁴⁴]. It was also observed that the impact toughness of the AZ80A Mg alloy joints had declined with the escalation in the diameter of the tool shoulder. This was due to the fact that the coarse grain structures formed at larger shoulder diameters had reduced the strengthening of the grain boundaries, which in turn had not improved the impact toughness to a greater level [¹⁸, ⁵¹].

4.5. Statistical influence of process parameters

ANOVA results described in the Table 5 indicate that the developed quadratic model for UTS is statistically significant (F = 19.57, p = 0.0022) with a high coefficient of determination (R² = 0.9724, Adjusted R² = 0.9227). Among the linear terms, traverse speed (T) exhibits the highest F-value (20.65, p = 0.0061) and a contribution of 11.40%, confirming it as the most influential primary factor affecting tensile strength. Although shoulder diameter (S) is also significant (F = 8.54, p = 0.0329; contribution = 4.72%), rotational speed (R) remains statistically insignificant (p = 0.1260). Importantly, the quadratic term T² shows a very high F-value (40.49, p = 0.0014) with a contribution of 22.36%, while S² contributes 28.64% (F = 51.86, p = 0.0008). The significance of T² confirms the nonlinear influence of traverse speed on tensile performance, indicating the existence of an optimal processing window. The interaction term TR is statistically significant (F = 8.91, p = 0.0307), whereas TS and RS are not (p > 0.05). Similarly, for impact toughness (as seen in the Table 6), the model is significant (F = 23.85, p = 0.0014; R² = 0.9772, Adjusted R² = 0.9363). Traverse speed dominates overwhelmingly with an F-value of 88.07 (p = 0.0002) and a contribution of 40.10%, far exceeding rotational speed (5.65%) and shoulder diameter (6.05%). None of the interaction or quadratic terms are statistically significant for impact toughness (p > 0.05), reinforcing the primary linear dominance of traverse speed.

The statistical dominance of traverse speed can be physically justified. In FSW, effective heat input per unit length can be conceptually expressed as:

(7) $$Q\alpha \frac{\omega .Y}{v}$$

where v is traverse speed. Unlike rotational speed, traverse speed directly controls tool-material interaction time per unit weld length, simultaneously influencing heat accumulation, strain rate, and dynamic recrystallization (DRX) kinetics. The significance of T² for UTS confirms that both excessively low and excessively high traverse speeds are detrimental. Low traverse speed increases thermal dwell time, promoting grain growth, while high traverse speed reduces plasticization and DRX effectiveness. SEM analysis shows refined grains (3.6 ± 0.8 μm) under optimized traverse speed, whereas non-optimal conditions produce coarser grains (~9.8 ± 2.4 μm). Hall–Petch estimation predicts ~40 MPa strengthening due to grain refinement, aligning with improved UTS trends [⁷, ²¹]. Impact toughness trends mirror ANOVA findings: the strong 40.10% contribution of traverse speed corresponds to enhanced crack deflection and energy absorption in fine-grained structures. Thus, numerical ANOVA results, grain size quantification, hardness distribution, and fracture behavior collectively confirm that traverse speed is the governing thermo-mechanical parameter controlling joint integrity.

4.6. Interactive impact of parameters

Three dimensional contour plots are graphical illustrations of the relationships prevailing amidst several independent process parameters and the outcomes, i.e., AZ80A Mg alloy joint’s impact toughness and tensile strength. These plots help us to understand visually, in an easier manner, how the parameters of the FSW process have interacted and impacted the friction stir welded joint’s properties [³¹]. Usually, in 3D contour plots, the self-reliant parameters were taken as the X & Y axis respectively and the outcome values (i.e., response variable) will be plotted on the Z axis. Three dimensional contour plots for the tensile strength of the AZ80A Mg alloy joints are illustrated in the Figure 8(a)–(c).

Figure 8
3D contour plots illustrating the interactive impact (a) traverse speed & rotational speed (b) shoulder diameter & traverse speed and (c) rotational speed & shoulder diameter on tensile strength of the friction stir welded AZ80A Mg alloy joints.

By examining these contour plots, it can be visualized that the fluctuation in the tensile strength was found to be more responsive to the modifications in the tool’s traverse speed than the modifications in the tool’s shoulder diameter and its rotational speed. At the same time, when the experiments were carried out at constant tool’s traverse speed, the shoulder diameter is found to impact the tensile strength in larger manner when compared with that of the rotational speed, as seen in the Figure 8(c). From these plots, it can also be understood that the interaction amidst the tool’s traverse speed and rotational speed had played a major role in impacting the tensile strength when compared with other interactions. This was mainly due to the fact that the volume of the frictional heat being generated is dependent on the traverse and rotational speed’s ratio and this heat in turn impacts the tensile strength of the fabricated joints [³⁹].

3D contour plots for the impact toughness of the AZ80A Mg alloy joints are portrayed in the Figure 9(a)–(c) and from these contour plots, it can be understood that the impact hardness is highly influenced by the tool’s traverse speed when compared with that of the tool’s shoulder diameter and rotational speed. Moreover, by comparing these contour plots, it can be visualized that the interaction amidst the tool’s shoulder diameter and the traverse speed had played a dominant role in impacting the impact toughness of the fabricated joints.

Figure 9
3D contour plots illustrating the interactive impact (a) traverse speed & rotational speed (b) shoulder diameter & traverse speed and (c) rotational speed & shoulder diameter on impact toughness of the friction stir welded AZ80A Mg alloy joints.

Further, by examining the microstructure of the fabricated joints, it can be understood that the interactions between shoulder diameter and traverse speed had impacted the rate of cooling and formation of joint bead. For instance, at constant rotational speed, employment of higher traverse speed and smaller shoulder diameter had resulted in improper mixing of plasticized metal, forming voids in the nugget zone and the increased effects of strain hardening had led to brittle weldments & had reduced the impact toughness of the joints [³¹, ³⁶].

4.7. Microstructural characterization of weld zones

With the objective of understanding the microstructural evolution occurred in the friction stir welded AZ80A Mg alloy, detailed scanning electron microscopy (SEM) examinations were carried out across different regions of the welded joint. The SEM micrographs representing the base metal (i.e., BM), heat affected zone (i.e., HAZ), thermo-mechanically affected zone (i.e., TMAZ), and stir zone (i.e., SZ) are illustrated in the Figure 10 (a) – (d) respectively.

Figure 10
SEM micrographs illustrating (a) base metal (b) heat affected zone (c) thermo-mechanically affected zone and (d) stir zone respectively.

The base metal (BM) illustrated in the Figure 10(a) exhibits the original microstructure of the AZ80A Mg alloy characterized by relatively coarse grains and the presence of intermetallic phases distributed along the grain boundaries. Since this region was unaffected by FSW, the microstructure had retained its as-received morphology. Adjacent to the parent metal lies the HAZ, which had experienced only thermal exposure during the welding process without significant plastic deformation [²³, ⁴²]. Due to the thermal cycle imposed by frictional heat generation, slight grain coarsening was observed in this region as seen in the Figure 10(b). The coarsened grains had led to localized softening, which had influenced the mechanical performance of AZ80A Mg alloy joint.

The thermo-mechanically affected zone (TMAZ) was subjected to both elevated temperatures and moderate plastic deformation. As a result, the grains in this region appear elongated and distorted as visible in the Figure 10(c) indicating partial plastic flow but insufficient strain to initiate complete dynamic recrystallization. The microstructure in the TMAZ therefore shows a transitional morphology between the HAZ and the stir zone. At the center of the weld lies the stir zone (SZ), also known as the nugget zone. This region had experienced the most severe plastic deformation and highest temperature due to the combined action of the rotating tool and frictional heat. Consequently, dynamic recrystallization occurred, producing fine equi-axed grains that are significantly smaller than those observed in the base metal. This refined microstructure visible in the Figure 10(d) had increased the grain boundary density and contributed towards improved resistance against crack propagation, thereby enhancing the mechanical properties of the welded joint. The observed microstructural variations across the weld zones confirms that the mechanical behavior of friction stir welded AZ80A Mg alloy joints was strongly influenced by the combined effects of thermal exposure, plastic deformation, and dynamic recrystallization during FSW [⁴⁵].

4.8. XRD phase analysis

The phase constitution of SZ was examined using X-ray diffraction (XRD) analysis to understand the microstructural changes induced by FSW in the AZ80A Mg alloy. Attained diffraction pattern i.e., XRD graphs illustrated in the Figure 11 predominantly exhibits strong peaks corresponding to α-Mg matrix, indicating that Mg had remained the dominant crystalline phase after FSW. In addition to Mg matrix peaks, only very weak or negligible diffraction signatures corresponding to the MgZn₂ intermetallic phase were observed. The absence or significant reduction in intensity of MgZn₂ peaks suggests that the originally present strengthening precipitates had undergone partial or near-complete dissolution during FSW [⁸, ²¹]. This observation was found to be consistent with the thermal cycle experienced during FSW. SZ was subjected to severe plastic deformation combined with elevated temperatures, often approaching 0.6–0.8 of the melting temperature of Mg alloys, which had promoted the dissolution of metastable and equilibrium precipitates such as MgZn₂. The EDS compositional analysis further supports this interpretation. As shown in Table of the XRD graph, SZ was primarily composed of Mg (69.28 wt%), with comparatively small fractions of Al (6.35 wt%), Zn (0.53 wt%), and Mn (0.15 wt%). The extremely low Zn content detected in the analyzed region indicates that Zn atoms had been predominantly retained in solid solution within Mg lattice rather than existing as discrete MgZn₂ precipitates, thereby corroborating the observations from the XRD analysis. Presence of oxygen and carbon had attributed mainly to surface oxidation and environmental contamination during sample preparation [¹²].

Figure 11
XRD analysis of SZ showing strong α-Mg diffraction peaks and the absence of significant MgZn2 intermetallic reflections, suggesting heat-induced precipitate dissolution and formation of a Mg-rich solid solution matrix.

The dissolution of MgZn₂ precipitates during excessive heat exposure had generated significant microstructural consequences. MgZn₂ particles normally act as microstructural strengthening sites and vacancy sinks. When these precipitates had dissolved under high thermal input, the redistribution of solute atoms and the disappearance of these second-phase particles had led to localized lattice instability and vacancy accumulation [¹⁶, ³⁰]. Furthermore, during subsequent cooling, the rapid solid-state diffusion and thermal contraction mismatch had promoted micro-void nucleation at previously precipitate-rich regions or solute segregation sites. Such void nucleation was further intensified by the intense plastic deformation occurring in the stir zone, which introduced a high density of dislocations and vacancy clusters. Thus, the combined XRD phase analysis and EDS quantitative composition results provide consistent evidence supporting microstructural evidence for the heat-induced dissolution of MgZn₂ precipitates and the associated micro-void formation mechanism in SZ.

5. OPTIMIZATION AND VALIDATION

5.1. Optimization of process parameters

The process of interlinking the impact toughness and tensile strength have to be formulated as the escalation in the tensile strength of the joints will generally lead to the deterioration of the impact toughness of the joints and hence there prevails a need for investigating both the mechanical properties together. Usually after formulating the empirical correlation and being verified for sufficiency, the criteria for optimizing the process parameters was fixed for determining the most ideal welding scenarios [³⁷, ⁵²]. In this research work, both graphical as well as statistical (i.e., mathematical) optimization techniques were employed by choosing the desirable goals for each & every parameter being taken into account and for the experimental outcomes.

5.2. Statistical optimization

In this work, the statistical optimization of the parameters of the FSW process was done using the Design expert 13 software. The statistical component of this software employs downhill model investigation algorithm and identifies one or more points in the domain of factors, which will maximize the objective functions. Then, the optimization segment of this software explores the ideal combination of factorial levels that will concurrently fulfill the requirements being placed (i.e., criteria related to optimization) for each & every experimental outcomes and parameter factors (i.e., optimization of multiple responses). This process of optimizing the multiple responses comprises of merging the individual goals into a comprehensive desirability function. This type of approach was desired owing to its flexibility in providing importance for individual experimental outcomes, giving equal weightage to all the responses, its ease of availability and simplicity [³⁷]. This approach comprises of revamping each & every predicted outcome, Z_i into a unit less entity restricted within 0 < fi < 1, where larger value of fi implies that the outcome value Z_i is more preferable. If fi = 0, it denotes an entirely unattractive outcome.

In this research work, the specific feasibility of each & every response, namely fi was computed based on the Equations 6–10. The structure of the function of desirability can be modified for each & every objective through the field of weightage, i.e., wfi. Weightage entity was employed for giving more attention to the lower & upper bounds or for focusing towards the objective value. In the feasibility objective function (i.e., F) each & every outcome was assigned a significance (s), proportionate to the other experimental outcomes, such that their significance varies from the minimal significant value, namely from 1. In this work, the experimental outcomes, namely impact toughness and tensile strength are converted into suitable feasibility scales based on the Equations 6–10. When the goal is to attain maximum, the feasibility was described by

(8) $$f_i=\{\begin{array}{l}0\\ Z_i<Smal{l}_i\\ {\left(\frac{Z_i-Smal{l}_i}{Larg{e}_i-Smal{l}_i}\right)}^{w{f}_i},Smal{l}_i<Y_i<Hig{h}_i\\ Z_i<Hig{h}_i\\ 1\end{array}$$

When the goal is to attain minimum, the feasibility was described by

(9) $$f_i=\{\begin{array}{l}0\\ Z_i<Smal{l}_i\\ {\left(\frac{Larg{e}_i-Z_i}{Larg{e}_i-Smal{l}_i}\right)}^{w{f}_i},Smal{l}_i<Y_i<Hig{h}_i\\ Z_i<Hig{h}_i\\ 1\end{array}$$

When the goal is to attain the target, the feasibility was described by

(10) $$f_i=\{\begin{array}{l}{\left(\frac{Z_i-Smal{l}_i}{T_i-Smal{l}_i}\right)}^{wf{1}_i},Smal{l}_i<Y_i<Hig{h}_i\\ {\left(\frac{Z_i-Larg{e}_i}{T_i-Larg{e}_i}\right)}^{wf{2}_i},T_i<Y_i<Hig{h}_i\\ 0,Otherwise\end{array}$$

When the goal is within the spectrum, the feasibility was described by

(11) $$f_i=\{\begin{array}{l}1,Smal{l}_i<Y_i<Hig{h}_i\\ 0,Otherwise\end{array}$$

(12) $$F={\left({\prod }_{n-1}^n{f}_i^{S_i}\right)}^{\frac{1}{\sum r_i}}$$

In this research work, a total of 5 distinct factors were taken into account by selecting the desired objectives for each & every factor and for every experimental outcome and these details are described in the Table 7.

Table 7
Criteria employed for statistical optimization.

FACTORS/OUTCOMES	CATEGORY	GOAL	LOWER LIMIT	UPPER LIMIT
Traverse speed (T)	Factor	In range	0.65 mm/sec	1.35 mm/sec
Shoulder diameter (S)	Factor	In range	11.4 mm	15.6 mm
Rotational speed (R)	Factor	In range	896 rpm	1603 rpm
Impact toughness	Outcome	Maximize	16.3 J	24 J
Tensile strength	Outcome	Maximize	133.5 MPa	171 MPa
Impact toughness and tensile strength	Outcome	Maximize both outcomes	Impact toughness: 16.3 J, Tensile strength: 133.5 MPa	Impact toughness: 24 J, Tensile strength: 171 MPa
Impact toughness and tensile strength	Outcome	Target tensile strength (171 MPa) and maximize impact toughness	Impact toughness: 16.3 J, Tensile strength: 171 MPa	Impact toughness: 24 J, Tensile strength: 171 MPa
Impact toughness and tensile strength	Outcome	Target impact toughness (24 J) and maximize tensile strength	Impact toughness: 24 J, Tensile strength: 133.5 MPa	Impact toughness: 24 J, Tensile strength: 171 MPa

The optimization plots being attained for the criteria described in the Table 7 are illustrated in the Figure 12(a)–(e).

Figure 12
Optimizing plots for (a) Maximizing impact toughness (b) Maximizing tensile strength (c) Maximizing impact toughness & targeting parent metal’s tensile strength (d) Maximizing tensile strength & targeting parent metal’s impact toughness and (e) Maximizing both impact toughness & tensile strength.

The relevant optimized process parameters and experimental outcomes are described in the Table 8.

Table 8
Optimized parameters and their outcomes.

FACTORS/OUTCOMES	TRAVERSE SPEED (T), MM/SEC	SHOULDER DIAMETER (S), MM	ROTATIONAL SPEED (R), RPM	PREDICTED TENSILE STRENGTH (MPA)	PREDICTED IMPACT TOUGHNESS (J)	COMPOSITE DESIRABILITY (D)
Maximizing impact toughness	1003	1.25	12.33	169.85	24.00	0.972
Maximizing tensile strength	1074	1.21	13.03	171.00	23.12	0.964
Targeting impact toughness (24 J) and Maximizing tensile strength	925	1.35	12.06	168.92	24.00	0.951
Targeting tensile strength (171 MPa) and Maximizing impact toughness	1060	1.239	12.87	171.00	23.78	0.968
Maximizing both tensile strength and impact toughness (Multi Objective)	1021	1.246	12.868	170.92	23.85	0.981

5.3. Diagrammatic optimization

The statistical optimization helped to locate one or more positions for maximizing the experimental outcomes [⁵³, ⁵⁴]. At the same time, in order to optimize the parameters associated with multiple experimental outcomes employing graphical diagrams, zones have to be defined and these zones are the places where the prerequisite concurrently meet the formulated criterions. Contour plots can be categorized as the outcome of overlaying or superimposing specific outcome curves and with the help of these plots, search for ideal combination can be attained in a visual manner [³¹, ⁵⁵].

The exact location of the viable values of experimental outcomes in the factor span is displayed in a clear manner during the employment of graphical or diagrammatic optimization. During this optimization methodology, the locations not fitting with the optimization scenarios are shaded as seen in the Figure 13(a)–(c). The processing charts for the friction stir welded of AZ80A Mg alloy joints are displayed in the Figure 13(a)–(c).

Figure 13
Processing charts for the friction stir welded AZ80A Mg alloy joints (a) shoulder diameter & traverse speed (b) shoulder diameter & rotational speed and (c) traverse speed & rotational speed.

It can be observed in these charts that for each & every outcome, both the upper and lower limits were selected based on the results attained from statistical optimization. 5^th criterion being put forward in the statistical optimization was also taken into consideration in this diagrammatic optimization processes and was used for generating the overlay plots.

5.4. Experimental validation

In order to verify the adequacy and predictive capability of the developed RSM models, confirmation experiments were conducted using the optimized process parameters predicted by the statistical models. Three confirmatory experimental runs were carried out under welding conditions selected randomly within the optimized parameter domain. For each confirmation run, the friction stir welding (FSW) experiments were performed using the specified combinations of traverse speed, rotational speed, and shoulder diameter obtained from the optimization results. To improve the reliability of the validation process, three specimens were extracted from each welded joint, and the mechanical tests were performed to determine ultimate tensile strength and impact toughness. The average value of the three experimental measurements was considered as the representative response for each confirmatory experiment [⁵⁶].

The predicted values obtained from the developed regression models were then compared with the experimentally measured values, and the corresponding percentage error was calculated. The comparison between predicted and experimental responses is presented in Table 9. The results indicate that the percentage error between predicted and experimental values is very small, ranging from 0.30% to 0.33% for tensile strength and 1.75% to 2.08% for impact toughness. These deviations fall well within the acceptable limits for empirical modeling and demonstrate a strong agreement between the predicted and experimental results [⁵⁷].

Table 9
Confirmatory experiments carried out employing optimized values of process parameters.

NO	TRAVERSE SPEED (MM/SEC)	ROTATIONAL SPEED (RPM)	SHOULDER DIAMETER (MM)	OUTCOMES	PREDICTED VALUE	REALISTIC EXPERIMENTAL VALUE	PERCENTAGE OF ERROR
1	1.25	1003	12.33	Impact Toughness, J	24.00	23.58	1.75
				Tensile Strength, MPa	169.85	169.32	0.31
2	1.21	1074	13.03	Impact Toughness, J	23.12	22.64	2.08
				Tensile Strength, MPa	171.00	170.48	0.30
3	1.246	1021	12.868	Impact Toughness, J	23.85	23.41	1.84
				Tensile Strength, MPa	170.92	170.36	0.33

5.5. Inferences from SEM fractography

Tensile specimens were machined according to ASTM E8/E8M (Sub-size specimen) standards. The specimens were extracted transverse to the welding direction to evaluate joint strength across the weld interface. The weld nugget was positioned at the center of the gauge length to ensure that fracture, if any, occurred within the weld or heat-affected region. Gauge length of the tensile specimen was about 25mm, along with a width of 6mm, 6mm thickness and overall length of about 100mm. Charpy V-notch impact testing was conducted as per ASTM E23. The notch was positioned at the center of the weld nugget, with the crack propagation direction oriented perpendicular to the welding direction.

Tensile specimen extracted from the AZ80A Mg alloy joints fabricated during the confirmatory experiments described in the Table 8 were subjected to tensile test and the fractured surfaces of these tensile specimen were subjected to scanning electron microscopy (SEM) analysis. The basic objective of carrying out this SEM analysis of the fractured surfaces is to gain better understanding about the mechanism of fracture and deformation nature of the AZ80A Mg alloy joints, so as to confirm the impact of the chosen process parameters on the integrity of mechanical properties [¹⁴, ⁴⁴]. From the SEM images of the fractured surfaces illustrated in the Figure 14(a), existence of uniformly dispersed, fine dimples on the surface of fracture, reveals us that the specimen extracted from the joint fabricated under optimized process parameter scenarios had undergone a ductile category of fracture.

Figure 14
SEM images of the tensile fractured surface of specimen extracted from the AZ80A Mg alloy joints fabricated under optimized parameter values taken at two distinctive magnifications (a) 250X and (b) 500X.

Moreover, the presence of micro-void amalgamation reveals us the occurrence of reasonable plastic deformation before encountering fracture, implying that the employment of optimized process parameters had made its vital contribution towards structural integrity and superior toughness of the AZ80A Mg alloy joints. Employing optimized shoulder diameter of 12.868 mm along with 1.246 mm/sec traverse speed had promoted efficient flow of the plasticized metal, leading to uniform refinement of grains in the nugget zone and this in turn had contributed for the finer, deeper dimples on the fractured surfaces [³³]. At the same time, the optimized rotational speed of 1021 rpm had paved path for generation friction heat in ideal volumes, thereby eliminating lavish growth of grains and inheriting the mechanical performance of the joints. In addition to this, the non-appearance of large crevices confirmed the flaw free attribute of the joint, revealing the effectiveness of the employed multi-objective optimization strategy.

Presence of uniformly dispersed, fine dimples illustrated in the Figure 14(b), confirms us the improved impact toughness and ductility of the friction stir welded AZ80A Mg alloy joints and can be related with the 23.41 J impact toughness attained during the employment of optimized process parameter values. In addition to this, the microstructural uniformity observed in the SEM images correlates with the achieved tensile strength of approximately 170.48 MPa and suggests that the optimized process parameters contributed to minimizing visible welding defects and promoting improved structural integrity

6. CONCLUSIONS

In this research work, an attempt was put forward to enhance the tensile strength and impact hardness of the friction stir welded AZ80A Mg alloy plates by employing two distinctive optimization strategies namely DoE and RSM combined together with statistical and diagrammatic multi objective optimization strategies. FSW parameters namely tool’s traverse speed, its speed of rotation and its shoulder diameter were taken into consideration and the significant results of this research work are described below:

• Tool’s traverse speed was found be the most significant process parameters influencing the tensile strength and impact toughness of the friction stir welded AZ80A Mg alloy plates, followed by the tool’s shoulder diameter and its rotational speed.

• From the 3D contour plots, it was revealed that the interaction amidst the tool’s traverse speed and rotational speed had played a major role in impacting the tensile strength when compared with other interactions.

• Likewise, from the generated 3D contour plots, it was visualized that the interaction amidst the tool’s shoulder diameter and the traverse speed had played a dominant role in impacting the impact toughness of the fabricated joints.

• Employment of larger tool shoulder diameters (greater than 13.5 mm) widened the heat impacted and thermo-mechanically impacted zone, thereby deteriorating the tensile strength of the AZ80A Mg alloy joints. Similarly, impact toughness of the AZ80A Mg alloy joints declined with the escalation in the shoulder diameter.

• Tool traverse speed of 1.246 mm/sec along with a 12.868 mm tool shoulder diameter and rotational speed of 1021 rpm were found to be the optimized process parameter combinations which led to the generation of flaw free AZ80A Mg alloy joints, which exhibited an impact toughness of 23.41 J and tensile strength of 170.48 MPa.

• Results of confirmatory experimental runs conducted by employing the optimized process parameter values revealed the presence of negligible discrepancy amidst the anticipated and realistic experimental values, which in turn confirmed the perfect accuracy of the formulated statistical model.

• SEM analysis of the tensile fractured surfaces of the AZ80A Mg alloy joints fabricated under the optimized process parameter conditions exhibited a ductile category of failure, identified by fine & small dimples, micro-void coalescence, thereby revealed joint’s superior capability w.r.t plastic deformation.

7. BIBLIOGRAPHY

• ^[1] HAN, S.H., KANG, C.G., SUNG, J.H., “Investigation of viscosity properties for rheology forming of AM50A magnesium alloy”, Journal of Materials Processing Technology, v. 187–188, pp. 335–338, 2007. doi: http://doi.org/10.1016/j.jmatprotec.2006.11.195.
  » https://doi.org/10.1016/j.jmatprotec.2006.11.195
• ^[2] QIAO, X., PAN, B., LIU, S., et al., “Thermal fatigue behavior and microstructure evaluation of Mg alloys after precipitation hardening”, Journal of the Minerals Metals & Materials Society, v. 75, n. 12, pp. 5819–5832, 2023. doi: http://doi.org/10.1007/s11837-023-06189-4.
  » https://doi.org/10.1007/s11837-023-06189-4
• ^[3] WANG, G.G., WEILER, J.P., “Recent developments in high-pressure die-cast magnesium alloys for automotive and future applications”, Journal of Magnesium and Alloys, v. 11, n. 1, pp. 78–87, 2023. doi: http://doi.org/10.1016/j.jma.2022.10.001.
  » https://doi.org/10.1016/j.jma.2022.10.001
• ^[4] JIANG, J.F., WANG, Y., QU, J.J., “Effect of process parameters on microstructure and properties of AM50A magnesium alloy parts formed by double control forming”, Transactions of Nonferrous Metals Society of China, v. 24, n. 2, pp. 1630–1637, 2010.
• ^[5] ZHAO, D., ZHAO, C., CHEN, X., et al., “Compressive deformation of as-extruded LPSO-containing Mg alloys at different temperatures”, Journal of Materials Research and Technology, v. 16, pp. 944–959, 2022. doi: http://doi.org/10.1016/j.jmrt.2021.12.053.
  » https://doi.org/10.1016/j.jmrt.2021.12.053
• ^[6] XIE, H., ZHAO, H., GUO, X., et al., “Recent progress on cast magnesium alloy and components”, Journal of Materials Science, v. 59, n. 22, pp. 9969–10002, 2024. doi: http://doi.org/10.1007/s10853-024-09459-x.
  » https://doi.org/10.1007/s10853-024-09459-x
• ^[7] GUO, B., CONG, X., DU, W., et al., “Effect of AC signal ratio on the formation, microstructure and mechanical properties of welded joint of AZ31B Mg alloy during the AC/DC mix GTAW”, Journal of Manufacturing Processes, v. 85, pp. 306–313, 2023. doi: http://doi.org/10.1016/j.jmapro.2022.11.062.
  » https://doi.org/10.1016/j.jmapro.2022.11.062
• ^[8] LI, J., BAI, S., DONG, Z., et al., “Research on extrusion welding of dissimilar magnesium alloys: Mg–Al–Zn and Mg–Gd”, Journal of Materials Research and Technology, v. 35, pp. 1296–1310, 2025. doi: http://doi.org/10.1016/j.jmrt.2025.01.103.
  » https://doi.org/10.1016/j.jmrt.2025.01.103
• ^[9] OUYANG, S., WANG, F., SHE, J., et al., “Recent advancements in the design of micro/nanostructured superhydrophobic surfaces on magnesium alloys”, Journal of Magnesium and Alloys, v. 13, n. 2, pp. 457–485, 2025. doi: http://doi.org/10.1016/j.jma.2025.01.013.
  » https://doi.org/10.1016/j.jma.2025.01.013
• ^[10] BERTOLINI, R., SIMONETTO, E., GHIOTTI, A., et al., “Effects of metal surface treatment on the interfacial strength in magnesium alloy-based fiber metal laminates”, Procedia CIRP, v. 131, pp. 32–36, 2025. doi: http://doi.org/10.1016/j.procir.2024.09.008.
  » https://doi.org/10.1016/j.procir.2024.09.008
• ^[11] YELAMASETTI, B., ASLESHA, C.L., SARATHI, T.V., et al., “Synergistic effects of optimizing stir casting and friction stir welding for WC and Al2O3 based nano-particulates reinforced AA2014 and AA6061 alloys: Mechanical, weldability, corrosion, and microstructural morphological insights”, Journal of Materials Research and Technology, v. 35, pp. 4895–4909, 2025. doi: http://doi.org/10.1016/j.jmrt.2025.02.140.
  » https://doi.org/10.1016/j.jmrt.2025.02.140
• ^[12] Luk’yanova, E.A., TARYTINA, I.E., DOBATKINA, T.V., et al., “Prospects of development of new high-strength magnesium-based alloys”, Metal Science and Heat Treatment, v. 66, n. 7–8, pp. 498–503, 2024. doi: http://doi.org/10.1007/s11041-024-01077-3.
  » https://doi.org/10.1007/s11041-024-01077-3
• ^[13] SINGH, V.P., KUMAR, A., KUMAR, R., et al., “Effect of rotational speed on mechanical, microstructure, and residual stress behaviour of AA6061-T6 alloy joints through friction stir welding”, Journal of Materials Engineering and Performance, v. 33, n. 8, pp. 3706–3721, 2024. doi: http://doi.org/10.1007/s11665-023-08527-8.
  » https://doi.org/10.1007/s11665-023-08527-8
• ^[14] MEHDI, H., JAIN, S., SALAH, A.N., et al., “Effect of friction stir welding parameters on microstructure and mechanical properties of the dissimilar alloys of AZ91D and AA7075”, Proceedings of the Institution of Mechanical Engineers. Part E, Journal of Process Mechanical Engineering, v. 239, n. 4, pp. 2050–2063, 2025. doi: http://doi.org/10.1177/09544089231195444.
  » https://doi.org/10.1177/09544089231195444
• ^[15] KURITSYN, D.N., SILUYANOVA, M.V., KURITSYNA, V.V., “Friction stir welding of heat-resistant materials for aircraft engines”, Russian Engineering Research, v. 43, n. 7, pp. 857–861, 2023. doi: http://doi.org/10.3103/S1068798X23070158.
  » https://doi.org/10.3103/S1068798X23070158
• ^[16] YILDIZ, M., OZTURK, F., SHEIKH-AHMAD, J., “A comprehensive review on friction stir welding of high-density polyethylene”, Arabian Journal for Science and Engineering, v. 48, n. 9, pp. 11167–11210, 2023. doi: http://doi.org/10.1007/s13369-023-08048-5.
  » https://doi.org/10.1007/s13369-023-08048-5
• ^[17] LU, Y.J., LIU, Q.L., LIU, Y.D., et al., “Mechanical property and fracture characteristics of carbon fiber reinforced aluminum matrix composites prepared by friction stir welding”, Journal of Materials Research and Technology, v. 34, pp. 1080–1089, 2025. doi: http://doi.org/10.1016/j.jmrt.2024.12.108.
  » https://doi.org/10.1016/j.jmrt.2024.12.108
• ^[18] SENGUPTA, K., MONDAL, A.K., BOSE, D., et al., “Experimental investigation on metallurgical properties on effect of dual hybridization by ultrasonic and electric energy in friction stir welding for joining Inconel 625 to stainless steel 304”, Transactions of the Indian Institute of Metals, v. 77, n. 4, pp. 941–953, 2024. doi: http://doi.org/10.1007/s12666-023-03179-7.
  » https://doi.org/10.1007/s12666-023-03179-7
• ^[19] TENG, L., LU, X., LUAN, Y., et al., “Predicting axial force in friction stir welding thick 2219 aluminum alloy plate”, International Journal of Advanced Manufacturing Technology, v. 126, n. 3–4, pp. 1025–1034, 2023. doi: http://doi.org/10.1007/s00170-023-11175-w.
  » https://doi.org/10.1007/s00170-023-11175-w
• ^[20] KESHARWANI, G.S., BILLA, K.K., VERMA, J., et al., “Influence of tool velocity ratio on force-torque and mechanical properties of friction stir welded 2050-T84 Al-Li alloy plates”, Physica Scripta, v. 99, n. 7, pp. 075008, 2024. doi: http://doi.org/10.1088/1402-4896/ad505f.
  » https://doi.org/10.1088/1402-4896/ad505f
• ^[21] ZUO, L., HAN, Y., SHAO, W., et al., “Revealing microstructure evolution and mechanical properties of Al/Cu joints during high-speed friction stir welding”, Journal of Materials Research and Technology, v. 33, pp. 9276–9288, 2024. doi: http://doi.org/10.1016/j.jmrt.2024.11.226.
  » https://doi.org/10.1016/j.jmrt.2024.11.226
• ^[22] ELSHEIKH, A.H., ELMILIGY, M., EL-KASSAS, A.M., “Optimization of joint strength in friction stir welded wood plastic composites using ANFIS and Cheetah Optimizer”, Journal of Materials Research and Technology, v. 34, pp. 2539–2552, 2025. doi: http://doi.org/10.1016/j.jmrt.2024.12.257.
  » https://doi.org/10.1016/j.jmrt.2024.12.257
• ^[23] KHALAF, H.I., AL-SABUR, R., DERAZKOLA, H.A., “Effect of number of tool shoulders on the quality of steel to magnesium alloy dissimilar friction stir welds”, Archives of Civil and Mechanical Engineering, v. 23, n. 2, pp. 125, 2023. doi: http://doi.org/10.1007/s43452-023-00673-z.
  » https://doi.org/10.1007/s43452-023-00673-z
• ^[24] MUSTAFA, S.E., RAI, R.N., FIROZ, R., “Enhancement of joint properties and reduction of intermetallics in FSW of highly dissimilar Al/Ti alloys”, Welding in the World, v. 67, pp. 1393–1410, 2023. doi: http://doi.org/10.1007/s40194-023-01493-8.
  » https://doi.org/10.1007/s40194-023-01493-8
• ^[25] KUMAR, S., SINHA, A., “Effect of process parameter on microstructure and mechanical properties of friction stir welded Al-Li alloy 2050: state of the art review”, Physica Scripta, v. 100, n. 7, pp. 072003, 2025. doi: http://doi.org/10.1088/1402-4896/ade5d5.
  » https://doi.org/10.1088/1402-4896/ade5d5
• ^[26] LU, X., LUAN, Y., QIAO, J., et al., “Optimization of process parameters in friction stir welding of 2219 aluminum alloy thick plate”, International Journal of Advanced Manufacturing Technology, v. 129, n. 9–10, pp. 4201–4215, 2023. doi: http://doi.org/10.1007/s00170-023-12577-6.
  » https://doi.org/10.1007/s00170-023-12577-6
• ^[27] BOUKRAA, M., CHEKIFI, T., LEBAAL, N., “Friction stir welding of aluminum using a multi-objective optimization approach based on both Taguchi method and grey relational analysis”, Experimental Techniques, v. 47, n. 3, pp. 603–617, 2023. doi: http://doi.org/10.1007/s40799-022-00573-6.
  » https://doi.org/10.1007/s40799-022-00573-6
• ^[28] DARWINS, A.K., LEWISE, K.A.S., FAHAD, M., et al., “Parametric optimization of friction stir welding of ZE42 using NSGA-II”, International Journal on Interactive Design and Manufacturing, v. 18, n. 5, pp. 3193–3205, 2024. doi: http://doi.org/10.1007/s12008-023-01480-9.
  » https://doi.org/10.1007/s12008-023-01480-9
• ^[29] SUNDAR, A.S., RADHIKA, N., KUMAR, A., “Optimizing tool offset in static shoulder friction stir welding of Al6061 to Ti6Al4V for enhanced joint integrity”, Journal of Materials Research and Technology, v. 33, pp. 1494–1503, 2024. doi: http://doi.org/10.1016/j.jmrt.2024.09.118.
  » https://doi.org/10.1016/j.jmrt.2024.09.118
• ^[30] KALITA, K., BURANDE, D., GHADAI, R.K., et al., “Finite element modelling, predictive modelling and optimization of metal inert gas, tungsten inert gas and friction stir welding processes: A comprehensive review”, Archives of Computational Methods in Engineering, v. 30, n. 1, pp. 271–299, 2023. doi: http://doi.org/10.1007/s11831-022-09797-6.
  » https://doi.org/10.1007/s11831-022-09797-6
• ^[31] ALTENHOF, W.J., RACZY, A., LAFRAMBOISE, M., et al., “Numerical simulation of AM50A magnesium alloy under large deformation”, International Journal of Impact Engineering, v. 30, n. 2, pp. 117–142, 2004. doi: http://doi.org/10.1016/S0734-743X(03)00060-5.
  » https://doi.org/10.1016/S0734-743X(03)00060-5
• ^[32] SONG, Q., WANG, H., WANG, W., et al., “Optimization of mechanical and surface properties of friction stir welded dissimilar joint of 2507 SDSS and 317L ASS by controlling process parameters”, Journal of Materials Research and Technology, v. 29, pp. 416–426, 2024. doi: http://doi.org/10.1016/j.jmrt.2024.01.074.
  » https://doi.org/10.1016/j.jmrt.2024.01.074
• ^[33] RUSKIN BRUCE, A.R., PREM KUMAR, P., ARUL, K., et al., “Experimental characteristics and optimization of friction stir welded AA5052-AA6061 using RSM technique”, Materials Today: Proceedings, v. 59, n. 2, pp. 1379–1387, 2022. doi: http://doi.org/10.1016/j.matpr.2021.12.099.
  » https://doi.org/10.1016/j.matpr.2021.12.099
• ^[34] JAIN, S., MISHRA, R.S., MEHDI, H., et al., “Optimization of processing variables of friction stir welded dissimilar composite joints of AA6061 and AA7075 using response surface methodology.”, Journal of Adhesion Science and Technology, v. 38, n. 6, pp. 949–968, 2024. doi: http://doi.org/10.1080/01694243.2023.2243682.
  » https://doi.org/10.1080/01694243.2023.2243682
• ^[35] HEIDARZADEH, A., MOTALLEB-NEJAD, P., BARENJI, R.V., et al., “The origin of the maximum hardness of the friction stir welded single-phase Cu-Zn plates: RSM, EBSD, and TEM investigation”, Materials Chemistry and Physics, v. 223, pp. 9–15, 2019. doi: http://doi.org/10.1016/j.matchemphys.2018.10.033.
  » https://doi.org/10.1016/j.matchemphys.2018.10.033
• ^[36] ELATHARASAN, G., KUMAR, V.S.S., “An experimental analysis and optimization of process parameter on friction stir welding of AA6061-T6 aluminum alloy using RSM”, Procedia Engineering, v. 64, pp. 1227–1234, 2013. doi: http://doi.org/10.1016/j.proeng.2013.09.202.
  » https://doi.org/10.1016/j.proeng.2013.09.202
• ^[37] NISHANT, JHA, S.K., PRAKASH, P., “Numerical analyses of underwater friction stir welding using computational fluid dynamics for dissimilar aluminum alloys”, Journal of Materials Engineering and Performance, v. 33, n. 22, pp. 12620–12637, 2024. doi: http://doi.org/10.1007/s11665-023-08824-2.
  » https://doi.org/10.1007/s11665-023-08824-2
• ^[38] DAS, P.P., CHAKRABORTY, S., “Optimization of friction stir welding processes using hybrid-Taguchi methods: A comparative analysis”, International Journal on Interactive Design and Manufacturing, v. 17, n. 3, pp. 1021–1038, 2023. doi: http://doi.org/10.1007/s12008-022-01017-6.
  » https://doi.org/10.1007/s12008-022-01017-6
• ^[39] NIRGUDE, S.K., KALPANDE, S.D., “Parametric optimization of friction stir welding of AA6101-T64 and pure Cu using response surface methodology”, International Journal on Interactive Design and Manufacturing, v. 18, n. 3, pp. 1789–1807, 2024. doi: http://doi.org/10.1007/s12008-023-01207-w.
  » https://doi.org/10.1007/s12008-023-01207-w
• ^[40] NGUYEN, T.T., NGUYEN, C.T., VAN, A.L., “Sustainability-based optimization of dissimilar friction stir welding parameters in terms of energy saving, product quality, and cost-effectiveness”, Neural Computing & Applications, v. 35, n. 7, pp. 5221–5249, 2023. doi: http://doi.org/10.1007/s00521-022-07898-8.
  » https://doi.org/10.1007/s00521-022-07898-8
• ^[41] MEHDI, H., BATRA, L., SINGH, A.P., et al., “Multi-response optimization of FSW process parameters of dissimilar aluminum alloys of AA2014 and AA6061 by response surface methodology (RSM)”, International Journal on Interactive Design and Manufacturing, v. 18, pp. 1–16, 2023.
• ^[42] KOLTYGIN, A.V., BAZHENOV, V.E., “Influence of the chemical composition and heat treatment modes on the phase composition and mechanical properties of the ZK51A (ML12) alloy”, Russian Journal of Non-Ferrous Metals, v. 59, n. 2, pp. 190–199, 2018. doi: http://doi.org/10.3103/S1067821218020049.
  » https://doi.org/10.3103/S1067821218020049
• ^[43] BINDHUSHREE, B.S., SEVVEL, P., SHANMUGANATAN, S.P., et al., “Influence of tool geometry on mechanical and microstructural characteristics of friction stir welded cast alloys”, Practical Metallography, v. 61, n. 5, pp. 292–318, 2024. doi: http://doi.org/10.1515/pm-2023-1055.
  » https://doi.org/10.1515/pm-2023-1055
• ^[44] MENG, S., LIU, H., XIAO, J., et al., “A method for process parameter optimization of simultaneous double-sided friction stir welding using a heat transfer model”, International Journal of Advanced Manufacturing Technology, v. 121, n. 5-6, pp. 3747–3758, 2022. doi: http://doi.org/10.1007/s00170-022-09544-y.
  » https://doi.org/10.1007/s00170-022-09544-y
• ^[45] HU, H., “Grain microstructure evolution of Mg (AM50A)/SiCp metal matrix composites”, Scripta Materialia, v. 39, n. 8, pp. 1015–1022, 1998. doi: http://doi.org/10.1016/S1359-6462(98)00281-4.
  » https://doi.org/10.1016/S1359-6462(98)00281-4
• ^[46] ZHAO, F., SHI, L., GAO, J., et al., “Roles of anisotropy and mutual transformation of the phases β/γ in multiphase-field simulation of IMCs in friction stir welding of Al/Mg alloys”, Journal of Materials Research and Technology, v. 33, pp. 2057–2070, 2024. doi: http://doi.org/10.1016/j.jmrt.2024.09.204.
  » https://doi.org/10.1016/j.jmrt.2024.09.204
• ^[47] LI, X., ZHAO, Q., SU, H., et al., “Intermetallic compounds formation in dissimilar friction stir welding of Mg/Cu alloys”, Acta Metallurgica Sinica, v. 37, n. 9, pp. 1523–1532, 2024. doi: http://doi.org/10.1007/s40195-024-01714-z.
  » https://doi.org/10.1007/s40195-024-01714-z
• ^[48] DERAZKOLA, H.A., KUBIT, A., “Effects of water cooling of friction stir welding of magnesium alloy stiffness joint”, Journal of Advanced Joining Processes, v. 10, pp. 100257, 2024. doi: http://doi.org/10.1016/j.jajp.2024.100257.
  » https://doi.org/10.1016/j.jajp.2024.100257
• ^[49] MEHDI, H., JAIN, S., MSOMI, V., et al., “Effect of Intermetallic Compounds on Mechanical and Microstructural Properties of Dissimilar Alloys Al-7Si/AZ91D”, Journal of Materials Engineering and Performance, v. 33, n. 10, pp. 4781–4793, 2024. doi: http://doi.org/10.1007/s11665-023-08302-9.
  » https://doi.org/10.1007/s11665-023-08302-9
• ^[50] LASKA, A., SADEGHI, B., SADEGHIAN, B., et al., “Temperature evolution, material flow, and resulting mechanical properties as a function of tool geometry during friction stir welding of AA6082”, Journal of Materials Engineering and Performance, v. 32, n. 23, pp. 10655–10668, 2023. doi: http://doi.org/10.1007/s11665-023-08671-1.
  » https://doi.org/10.1007/s11665-023-08671-1
• ^[51] ZHAI, M., SHI, L., SONG, H.T., et al., “Comparative numerical analysis between butt and lap joints of Mg alloy friction stir welding with considering tilted tool effect”, Journal of Manufacturing Processes, v. 120, pp. 951–965, 2024. doi: http://doi.org/10.1016/j.jmapro.2024.05.017.
  » https://doi.org/10.1016/j.jmapro.2024.05.017
• ^[52] ZHAO, F., SHI, L., GAO, J., et al., “Multiphase-field analysis of the intermetallic compounds formation and evolution in friction stir welding of dissimilar Al/Mg alloys”, Journal of Materials Research and Technology, v. 32, pp. 1206–1222, 2024. doi: http://doi.org/10.1016/j.jmrt.2024.07.193.
  » https://doi.org/10.1016/j.jmrt.2024.07.193
• ^[53] GUNASEKARAN, J., SEVVEL, P., JOHN SOLOMON, I., et al., “Multi objective optimization of parameters during FSW of AZ80A – AZ31B Mg alloys using grey relational analysis”, Journal of Mechanical Science and Technology, v. 38, n. 9, pp. 4971–4982, 2024. doi: http://doi.org/10.1007/s12206-024-0832-3.
  » https://doi.org/10.1007/s12206-024-0832-3
• ^[54] MYŚLIWIEC, P., SZAWARA, P., KUBIT, A., et al., “FSW Optimization”, Materials, v. 18, n. 2, pp. 448, 2025. doi: http://doi.org/10.3390/ma18020448. PubMed PMID: 39859919.
  » https://doi.org/10.3390/ma18020448
• ^[55] KUMAR, J., KUMAR, G., MEHDI, H., et al., “Optimization of FSW parameters on mechanical properties of different aluminum alloys of AA6082 and AA7050 by response surface methodology”, International Journal on Interactive Design and Manufacturing, v. 18, pp. 1–13, 2023.
• ^[56] WANG, Y., CHEN, Y., ZHOU, L., et al., “Improving the corrosion resistance of friction stir welding joint of 7050 Al alloy via optimizing the process route”, Journal of Materials Research and Technology, v. 24, pp. 8098–8108, 2023. doi: http://doi.org/10.1016/j.jmrt.2023.05.053.
  » https://doi.org/10.1016/j.jmrt.2023.05.053
• ^[57] MEHDI, H., MISHRA, R.S., “An experimental analysis and optimization of process parameters of AA6061 and AA7075 welded joint by TIG+FSP welding using RSM”, Advances in Materials and Processing Technologies, v. 8, n. 1, pp. 598–620, 2022. doi: http://doi.org/10.1080/2374068X.2020.1829952.
  » https://doi.org/10.1080/2374068X.2020.1829952

DATA AVAILABILITY

All data that support the findings of this study are included within the article (and any supplementary files).

Publication Dates

• Publication in this collection
  15 May 2026
• Date of issue
  2026

History

• Received
  19 Nov 2025
• Accepted
  15 Mar 2026