www.rsisinternational.org
Page 3081
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue VI, June 2026
Advancing Tilt Angle Control in Friction Stir Welding using Fuzzy
Logic
1,3
Mayowa Abioye,
2
Atakpa Noah Ojolimi,
3
Jonah Isaac
1
Department of Mechanical Engineering, University of Johannesburg, Johannesburg, South Africa.
2
National Space Research and Development Agency, Abuja, Nigeria.
3
Department of Mechanical Engineering, Kogi State Polytechnic, Lokoja, Nigeria.
DOI: https://doi.org/10.51583/IJLTEMAS.2026.150600225
Received: 12 July 2026; Accepted: 17 July 2026; Published: 25 July 2026
ABSTRACT
Friction Stir Welding (FSW) is a solid-state joining technique renowned for its ability to produce high-quality
welds in aluminum and other non-ferrous metals. A critical parameter influencing weld integrity is the tool tilt
angle, which, if improperly set, can lead to defects such as tunnel voids, surface irregularities, or poor material
mixing. This study presents a fuzzy logic-based control system for real-time adjustment of the tool tilt angle
during FSW. The proposed system utilizes inputs such as torque, temperature, and plunge depth deviation to
infer optimal tilt corrections using a Mamdani-type fuzzy inference system. Triangular membership functions
and an expert-defined rule base were implemented in MATLAB to simulate adaptive tilt adjustments. Simulation
results showed that for a torque input of 10 Nm, a temperature of 45 °C, and a plunge deviation of 0.5 mm, the
system produced a tilt adjustment value of 0.224°, as calculated through centroid defuzzification. Surface plots
confirmed that the controller responds smoothly to varying conditions, with tilt adjustments ranging from -0.4°
to +0.5° based on combined sensor inputs. The proposed system offers a robust and flexible solution suitable for
integration into CNC-based FSW platforms for improved automation and weld quality.
Keywords: Friction Stir Welding, Fuzzy Logic, Tilt Angle, Torque, Temperature, Plunge.
INTRODUCTION
Friction Stir Welding (FSW), developed by The Welding Institute (TWI) in 1991, has emerged as a revolutionary
solid-state joining technique, especially for materials such as aluminum and its alloys that are often challenging
to weld using traditional fusion-based methods[1]. Unlike conventional welding, FSW does not involve melting
of the base materials, resulting in superior mechanical properties, reduced residual stresses, and minimal
distortion[2][3].To further advance research of FSW ,Among the various process parameters that govern weld
quality in FSW such as tool rotational speed, traverse speed, axial force, and plunge depth the tilt angle of the
tool plays a critical role. It influences material flow, heat generation, forging action, and the consolidation of the
stirred material[4]. An optimal tilt angle ensures continuous contact between the tool shoulder and the workpiece,
promoting effective stirring and reducing surface or subsurface defects.
In conventional FSW operations, the tilt angle is typically set to a fixed value before welding begins and remains
unchanged throughout the process. However, this static approach may not be sufficient in practical applications
where process conditions vary due to differences in material thickness, joint configuration, thermal gradients, or
tool wear[5]. In such cases, a fixed tilt angle can result in inconsistent weld quality, surface irregularities, tunnel
voids, or incomplete bonding.
To address this limitation, adaptive control methods are gaining traction in FSW research. Fuzzy Logic Control
(FLC), known for its robustness in handling nonlinear, multivariable systems with uncertain or imprecise inputs,
presents a compelling solution[6][7]. By mimicking human reasoning and expert decision-making, FLC can
www.rsisinternational.org
Page 3082
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue VI, June 2026
dynamically interpret sensor data such as torque, temperature, and plunge deviation and infer the necessary
adjustments to the tilt angle in real time[8].
This study proposes the development and implementation of a fuzzy logic-based control system for real-time tilt
angle adjustment during Friction Stir Welding. The goal is to enhance weld consistency, reduce defect formation,
and improve process adaptability under varying welding conditions.
LITERATURE REVIEW
Friction Stir Welding (FSW) has gained significant traction in recent decades as a reliable and efficient solid-
state joining technique, particularly for aluminum and other low-melting-point materials. Initially developed by
The Welding Institute (TWI) in 1991, FSW avoids melting the base materials, thereby reducing defects such as
porosity and hot cracking that are common in conventional fusion welding methods[9][10]. Among the several
process parameters that influence FSW quality such as rotational speed, travel speed, plunge depth, and tilt angle,
the tilt angle plays a critical role in maintaining effective material flow, adequate forging action, and proper
consolidation of the weld [4].
While early FSW processes relied on fixed tilt angles throughout the welding process, researchers have
recognized that a static tilt angle can limit weld adaptability and lead to inconsistencies when dealing with
varying plate thicknesses, material properties, or joint geometries. Several studies have emphasized that
improper tilt angles can result in surface irregularities, tunnel defects, or insufficient material mixing, all of
which degrade weld quality[11][12]. Consequently, there is growing interest in real-time control mechanisms
capable of adjusting process parameters, including tilt angle, in response to in-process feedback.
Fuzzy logic controllers (FLCs) have emerged as powerful tools in handling the uncertainties and nonlinearities
inherent in welding processes. As a rule-based decision-making framework, fuzzy logic mimics human reasoning
by transforming vague or imprecise sensor inputs into control actions using linguistic rules[13][14]. Several
researchers have applied fuzzy logic to FSW parameter control. For instance, [5] employed fuzzy inference to
regulate tool speed based on real-time temperature data, demonstrating improvements in mechanical properties
and defect minimization. Similarly, studies by [15] integrated fuzzy systems into robotic FSW platforms to
dynamically adjust tool path and force based on resistance feedback.
The integration of torque, temperature, and plunge deviation as inputs into a fuzzy inference system offers a
robust method to dynamically control tilt angle. Torque provides insight into the resistance experienced by the
tool, often indicating changes in material hardness or improper plunge. Temperature reflects the thermal balance
of the process, crucial for maintaining proper softening and flow without overheating. Plunge deviation helps
detect variations in plate thickness or tool misalignment critical for maintaining weld consistency. Recent works,
such as those by [16], have highlighted the potential of multi-sensor fuzzy-based control systems for achieving
adaptive welding quality across different operational scenarios.
Despite the promise of fuzzy logic systems, most implementations still rely on static rule bases defined by
experts. There is a growing trend toward enhancing FLCs through hybrid approaches that incorporate machine
learning for rule extraction and optimization[17]. Additionally, while dynamic tilt angle control has been
discussed conceptually, its practical application, especially in robotic or CNC-based FSW platforms, remains
relatively underexplored in the literature.
This study contributes to the body of knowledge by designing and experimentally validating a fuzzy logic-based
system for real-time tilt angle correction in FSW. By integrating real-time feedback and a Mamdani-type fuzzy
inference structure, the system demonstrates improved adaptability, weld consistency, and quality under varying
process conditions.
2.2
www.rsisinternational.org
Page 3083
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue VI, June 2026
Friction Stir Welding for Space Manufacturing
Advancing the missions involving orbital assembly, satellite servicing, lunar base construction, and deep-space
exploration will depend on robust joining processes that can fabricate and repair metallic structures with minimal
human intervention. Among the available solid-state joining technologies, Friction Stir Welding (FSW) is
particularly well suited for in-space manufacturing because it eliminates the need for filler materials, shielding
gases, and molten metal, making it highly compatible with vacuum and microgravity environments. A
fundamental challenge in deploying FSW in space is maintaining an optimal tool tilt angle throughout the
welding process[4].
Advancing tilt angle control is therefore essential to achieving consistent, high-quality welds in space. Emerging
technologies such as artificial intelligence, fuzzy logic, machine learning, digital twins, and sensor fusion provide
powerful tools for developing adaptive tilt angle control systems. These intelligent approaches enable predictive
decision-making, autonomous process optimization, and real-time correction of deviations without human
intervention. Consequently, they improve weld quality, minimize defects, enhance structural reliability, and
increase the overall efficiency of in-space manufacturing operations [5]. The advancement of adaptive tilt angle
control in Friction Stir Welding represents a critical enabling technology for the next generation of autonomous
space manufacturing.
METHODOLOGY
System Architecture
The proposed control architecture for tilt angle adjustment in Friction Stir Welding (FSW) integrates a various
input parameter, a fuzzy inference system (FIS), and an actuated tilt adjustment mechanism as shown in figure
1. The input parameters are key process variables specifically, torque, temperature, and plunge depth deviation.
These inputs are crucial indicators of tool resistance, thermal conditions, and tool alignment, respectively, and
are fed into the FIS for intelligent control decisions. The input variables depict the membership value input and
output, and fuzzy defines rules formulated by human experts who have good knowledge of the FSW plant.
3.2 Fuzzy Inference System Design
The fuzzy inference system, designed using the Mamdani fuzzy logic model in MATLAB, interprets these sensor
signals to generate real-time tilt angle corrections. The system utilizes a combination of triangular membership
functions to fuzzy the input data into linguistic categories such as Low, Medium, and High. For instance, the
torque input is classified into three fuzzy sets representing varying levels of resistance encountered by the tool;
this helps the controller determine whether increased plunge or tilt is required.
Figure 1: Tilt Angle Control in Friction Stir Welding
www.rsisinternational.org
Page 3084
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue VI, June 2026
Likewise, temperature membership functions play a vital role in managing the thermal dynamics of the welding
process. These are also divided into Low, Medium, and High fuzzy sets, with overlapping boundaries to ensure
soft transitions between states. This fuzzy categorization enables the system to respond gradually to changes in
temperature, preventing abrupt corrections that could destabilize the process. In temperature-sensitive operations
like FSW, such gradual adaptation is critical to maintaining consistent material flow, avoiding overheating, and
ensuring structural integrity of the weld. Also, the third input, plunge deviation, represents the variation between
the actual and desired tool plunge depth. Its membership functions allow the system to recognize whether the
deviation is minor or significant, guiding corrective actions accordingly to prevent over-penetration or
incomplete joining.
Based on these inputs, the fuzzy logic controller applies a rule base developed from expert knowledge and
experimental analysis to infer the appropriate output. The output variable tilt angle adjustment is represented by
fuzzy sets such as Decrease, No Change, and Increase. The FIS performs defuzzification using the centroid
method to convert the fuzzy output into a crisp tilt correction value. This output is then used to drive an actuated
tilt mechanism, which dynamically adjusts the tool orientation during welding. This closed-loop system allows
the tool to adapt in real time to changing conditions, thereby reducing defects and improving weld consistency
across varying joint configurations and material properties.
Fuzzy logic systems utilize membership functions to model imprecise and gradually changing input variables in
a way that supports intelligent and adaptive decision-making. These membership functions define how input
values are categorized into linguistic variables such as Low, Medium, High, Increase, or Decrease. For instance,
in a process like Friction Stir Welding (FSW) or other precision control applications, various input variables
such as torque, temperature, plunge deviation, and tilt angle correction can be fuzzified to enable smooth control
actions.
Torque membership functions classify torque values into fuzzy sets like Low with Triangular MF → µ_low(x)
= trimf(x; 0, 0, 10), Medium with Triangular MF → µ_medium(x) = trimf(x; 5, 10, 15), and High with Triangular
MF → µ_high(x) = trimf(x; 10, 20, 20) , allowing the system to interpret resistance levels encountered during
welding. This helps adjust process parameters to maintain weld quality. Similarly, temperature membership
functions group readings into overlapping categories, providing a mechanism for gradual thermal control, which
is essential in avoiding abrupt shifts that could compromise system stability.
Figure 2: Torque membership functions
Plunge deviation membership functions monitor the variation between actual and desired plunge depth,
categorizing deviations as Low, Medium, or High. This classification is particularly useful in precision tasks
where maintaining consistent depth is crucial to structural integrity and process success.
www.rsisinternational.org
Page 3085
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue VI, June 2026
Figure 3: Temperature membership functions
Overall, fuzzy logic membership functions enhance the ability of control systems to handle real-world variability
by supporting soft transitions between states. This makes them highly effective in applications such as robotics,
smart manufacturing, and digital agriculture, where environmental and operational conditions are often uncertain
or continuously shifting.
Figure 4: Plunge deviation membership functions
The temperature input variable, measured in degrees Celsius, is critical in processes like Friction Stir Welding,
where both high and low temperatures can negatively affect material mixing or cause overheating. To handle
this, temperature is fuzzified into three linguistic categories: Low, Medium, and High. The Low temperature
membership function is defined as a triangular function with full membership at 30°C that decreases linearly to
zero at 45°C. The Medium temperature set peaks at 47°C, with membership gradually increasing from 40°C and
tapering off by 55°C. Finally, the High temperature membership function rises starting at 50°C and reaches full
membership at 60°C, representing conditions of potential overheating. The overlapping nature of these triangular
membership functions allows the control system to interpret temperatures smoothly, enabling gradual and stable
adjustments rather than abrupt changes, which is essential for maintaining weld quality and process stability.
www.rsisinternational.org
Page 3086
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue VI, June 2026
Finally, tilt adjustment membership functions, as illustrated in the accompanying figure, map tilt angle correction
values to fuzzy sets like Decrease, No Change, and Increase. These functions guide the system in correcting the
tool’s tilt angle gently and accurately, ensuring that optimal orientation is preserved during operation.
Figure 5: Tilt adjustment membership functions
The defuzzification process for this experiment modelling uses a centroid technique method. The Centroid
method is used to convert the output fuzzy set into a crisp tilt correction value. This gives a weighted average of
the area under the combined output fuzzy set to compute the final tilt angle correction.
Crisp output =
∫
𝒙.𝝁(𝒙)𝒅𝒙
∫
𝝁(𝒙)𝒅𝒙
𝑬𝒒𝒏 (𝟏)
RESULTS AND DISCUSSION
This section presents the outcomes of the fuzzy inference system simulation and interprets the implications of
the findings. The results are visualized using surface plots to demonstrate how the system responds to various
combinations of input variables, namely torque, temperature, and plunge deviation in determining the
appropriate tilt adjustment.
The discussion aims to provide insights into the logic embedded within the fuzzy controller and how it reflects
practical considerations in applications such as friction stir welding.
Simulation Results
The surface plots in figures 6 and 7 illustrate how a fuzzy inference system determines tilt adjustment based on
different combinations of input variables specifically, torque, temperature, and plunge deviation. In the figure 6,
the relationship between torque and temperature with respect to tilt adjustment is shown. As torque increases
from 0 to 20 Nm and temperature remains within a moderate range (approximately 30–40°C), the tilt adjustment
value rises significantly.
This suggests that the system interprets high torque combined with moderate temperatures as a condition that
requires positive tilt correction, likely to reduce tool resistance or enhance process stability in operations such
as friction stir welding.
www.rsisinternational.org
Page 3087
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue VI, June 2026
Figure 6:
Surface plot for Torque vs Temperature & Tilt Adjustment
While figure 7 depicts the relationship between plunge deviation and torque with respect to tilt adjustment is
visualized. The surface shows that when plunge deviation is high (approaching 1 mm), the tilt adjustment is
significantly negative. As the deviation decreases toward zero, the tilt adjustment shifts toward zero or slightly
positive values. Interestingly, torque appears to have a less pronounced effect in this scenario, indicating that
plunge deviation is the dominant factor influencing tilt correction in this rule set. The system likely responds this
way to prevent over-penetration or material distortion by reducing the tool’s tilt when excessive plunge is
detected.
Figure 7: Plunge Deviation vs Torque & Tilt Adjustment
Together, these plots highlight the adaptive nature of the fuzzy logic controller. By integrating multiple sensory
inputs and applying rule-based reasoning, the system generates smooth and intelligent tilt adjustments. This
approach is particularly useful in real-time control applications where conditions vary continuously, such as
welding automation and FSW applications.
www.rsisinternational.org
Page 3088
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue VI, June 2026
Furthermore, figure 8 displays a Fuzzy Logic Rule Viewer for a system named TiltAngleAdjustmentFIS, which
likely controls the tilt angle adjustment of a friction stir welding tool based on three input parameters: Torque,
Temperature, and Plunge Deviation. The current input values are Torque = 10, Temperature = 45, and Plunge
Deviation = 0.5. Each of these inputs is shown with associated membership functions for instance, the third input
(Plunge Deviation) shows triangular membership functions in yellow, and the red vertical lines indicate the
current crisp input values' positions within these fuzzy sets.
Figure 8: Fuzzy Logic Rule Viewer
The rightmost column represents the output variable, Tilt Adjustment, which is computed based on fuzzy
inference rules (three in this case). The overlapping blue and white areas in the output plot reflect the combined
output from the activated rules, and the red vertical line represents the final defuzzified value: approximately
2.24e-01 (or 0.224). This value suggests the amount by which the tool's tilt angle should be adjusted in response
to the given input conditions, as determined by the fuzzy logic system for smoother or more precise operation
during the welding process. The system can be integrated into CNC-based FSW machines or retrofitted into
existing setups with minor modifications.
CONCLUSION
This study confirms the feasibility and effectiveness of using a fuzzy logic-based control system for real-time
tilt angle adjustment in Friction Stir Welding (FSW). By integrating sensor feedback specifically torque,
temperature, and plunge deviation into a Mamdani-type fuzzy inference system, the controller enables dynamic
and intelligent tilt corrections that enhance weld consistency, reduce defects, and improve adaptability to varying
welding conditions. The results highlight the system’s potential for integration into automated or CNC-based
FSW platforms, providing a robust and responsive approach to weld quality control. Future work will explore
the incorporation of machine learning techniques to optimize the fuzzy rule base and extend the system’s
capabilities to handle multi-pass welding, dissimilar material joints, and more complex welding geometries.
REFERENCES
1. G. Sahu, P. K. Sen, R. Sharma, and S. Bohidar, “A Review on Analysis of Friction Stir Welding Process
of Steel,” IJSRST -Themed Sect. Eng. Technol., vol. 1, no. 5, pp. 61–64, 2015.
2. D. A. P. Prabhakar et al., “A Review of Optimization and Measurement Techniques of the Friction Stir
Welding (FSW) Process,” J. Manuf. Mater. Process., vol. 7, no. 5, 2023, doi: 10.3390/jmmp7050181.
www.rsisinternational.org
Page 3089
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue VI, June 2026
3. E. T. A. and S. A. Akinlabi, Friction stir welding of dissimilar metals. Woodhead Publishing Series in
Welding and Other Joining Technologies, 2014. doi: 10.1533/9780857094551.241.
4. A. H. Elsheikh, “Applications of machine learning in friction stir welding: Prediction of joint properties,
real-time control and tool failure diagnosis,” Eng. Appl. Artif. Intell., vol. 121, no. 105961, 2023.
5. T. A. Shehabeldeen, M. A. Elaziz, A. H. Elsheikh, and J. Zhou, “Modeling of friction stir welding process
using adaptive neuro-fuzzy inference system integrated with harris hawks optimizer,” J. Mater. Res.
Technol., vol. 8, no. 6, pp. 5882–5892, 2019, doi: 10.1016/j.jmrt.2019.09.060.
6. I. Erenoglu, I. Eksin, E. Yesil, and M. Guzelkaya, “An intelligent hybrid fuzzy PID controller,” 20th Eur.
Conf. Model. Simul. Model. Methodol. Simul. Key Technol. Acad. Ind. ECMS 2006, vol. 5, no. Cd, pp.
62–66, 2006, doi: 10.7148/2006-0062.
7. R. M. M. and I. S. Noah E El-Zathry*, Stephen Akinlabi, Wai Lok Woo, Vivek Patel, “Enhancing Friction
Stir-Based Techniques with Machine Learning: A Comprehensive Review,” Mach. Learn. Sci. Technol,
2025.
8. M. P. Iqbal, R. Jain, and S. K. Pal, “Numerical and experimental study on friction stir welding of
aluminum alloy pipe,” J. Mater. Process. Technol., vol. 274, no. June, p. 116258, 2019, doi:
10.1016/j.jmatprotec.2019.116258.
9. J. Backer, “Feedback Control of Robotic Friction Stir Welding,” University West SE-46186 Trollhättan
Sweden, 2014. [Online]. Available: https://www.diva-portal.org/smash/record.jsf?pid=diva2:706044
10. M. P. Mubiayi and E. T. Akinlabi, “An overview on friction stir spot welding of dissimilar materials,” in
Transactions on Engineering Technologies: World Congress on Engineering and Computer Science
2014, 2015, pp. 537–549. doi: 10.1007/978-94-017-7236-5_38.
11. S. Verma and V. Kumar, “An Overview of Friction Stir Welding Tools,” Adv. Manuf. Technol. Comput.
Mater. Process. Charact., vol. 5762, pp. 19–27, 2022, doi: 10.1201/9781003203681-4.
12. C. Q. Zhang, J. D. Robson, O. Ciuca, and P. B. Prangnell, “Microstructural characterization and
mechanical properties of high power ultrasonic spot welded aluminum alloy AA6111-TiAl6V4 dissimilar
joints,” Mater. Charact., vol. 97, pp. 83–91, 2014, doi: 10.1016/j.matchar.2014.09.001.
13. L. Hang and D. Kim, “Enhanced Model-Based Predictive Control System Based on Fuzzy Logic for
Maintaining Thermal Comfort in IoT Smart Space,” MDPI-Applied Sci., vol. 8, no. 1031, 2018, doi:
10.3390/app8071031.
14. T. Nasir, M. Asmael, Q. Zeeshan, and D. Solyali, “Applications of Machine Learning to Friction Stir
Welding Process Optimization,” J. Kejuruter., vol. 32, no. 2, pp. 171–186, 2020, doi: 10.17576/jkukm-
2020-32(2)-01.
15. E. Marliana, A. Wahjudi, L. Nurahmi, I. M. L. Batan, and G. Wei, “Optimizing the Tuning of Fuzzy-PID
Controllers for Motion Control of Friction Stir Welding Robots,” J. Robot. Control, vol. 5, no. 4, pp.
1002–1017, 2024, doi: 10.18196/jrc.v5i4.21697.
16. Y. Chen and M. Jean, “Multi-response optimization of friction stir welding using fuzzy-grey system,”
High Temp. Mater. Process., vol. 43, no. 1, 2024, doi: 10.1515/htmp-2024-0005.
17. K. Balachandar and R. Jegadeeshwaran, “Enhancing Friction Stir Welding: Quality Machine Learning
Based Friction Stir Welding Tool Condition Monitoring,” Int. Res. J. Multidiscip. Technovation, vol. 6,
no. 3, pp. 245–259, 2024, doi: 10.54392/irjmt24319.
