How much difference can the order of welding operations really make in the production of an automotive control arm?
In welded manufacturing, the answer is clear: the welding sequence can be a critical process variable affecting the final geometry and distortion behavior of a component.
As the number of weld locations increases, the number of possible welding sequences grows rapidly. Evaluating these alternatives one by one can require significant engineering time, simulation effort, and technical expertise.
As part of the R&D activities carried out at OSKİM, this challenge was approached from a different perspective:
Engineering simulation and artificial intelligence were brought together within the same decision-making process.
The resulting project, “Reduction of Distortion in Automotive Control Arms through Simulation- and AI-Assisted Welding Sequence Optimization,” provides a practical example of how artificial intelligence can move beyond theory and support the solution of a real manufacturing engineering problem.
The Challenge: Welding-Induced Distortion
During welding, a high amount of heat is locally introduced into the component.
While certain regions of the part are heated, other areas remain at lower temperatures. During the subsequent cooling process, these temperature gradients can generate residual stresses and geometric changes in the component.
The result?
The part may deviate from its intended design geometry.
For automotive components such as control arms, where dimensional accuracy, structural performance, and safety are critical, welding-induced distortion must be carefully managed throughout the manufacturing process.
The welding sequence is one of the key variables influencing this behavior. Even when the same weld locations are used, changing the order in which they are performed can alter the thermal distribution and, consequently, the distortion of the component.
The engineering question is therefore not simply:
“How should the component be welded?”
but rather:
“In which sequence should the welds be performed to minimize distortion?”
As the Number of Alternatives Grows, So Does the Engineering Challenge
When a component contains only a few welding operations, the number of possible alternatives may remain manageable.
However, as the number of weld locations increases, the number of possible sequences grows very quickly.
Producing every alternative physically is impractical, while evaluating every scenario individually through conventional engineering methods can require considerable time.
This is exactly where the approach developed at OSKİM comes into play.
The objective is not to replace engineering knowledge, but to identify the most promising alternatives more efficiently from a large number of possible welding sequences.
Simulation and Artificial Intelligence in the Same Workflow
In OSKİM’s approach, different welding sequence alternatives are analyzed in Simufact Welding.
The distortion and process results obtained from these simulations are then evaluated using an artificial intelligence model developed in-house at OSKİM.
This brings two complementary technologies together for the same engineering problem:
Physics-Based Engineering Simulation + Artificial Intelligence
The simulation provides insight into the physical behavior of the welding process, while the AI model supports faster evaluation of the relationships between the different alternatives and their outcomes.
The overall workflow can be summarized as follows:
Generate Welding Sequence Alternatives → Analyze Them Through Simulation → Evaluate the Results with AI → Identify Promising Alternatives → Support the Engineering Decision
An important distinction should be made here.
Artificial intelligence is not positioned as an autonomous system that independently controls the welding process.
Instead, it functions as a decision-support tool that helps engineers evaluate alternatives faster and more systematically.
From 30 Working Days to 2–3 Working Days
One of the most significant outcomes of the project is the reduction in engineering lead time.
A welding sequence optimization study that could take approximately 30 working days using a conventional evaluation approach can be completed in approximately 2–3 working days with the simulation- and AI-assisted method developed by OSKİM.
This improvement is not limited to faster analysis.
It also creates significant potential for:
- evaluating a greater number of welding sequence alternatives,
- accelerating pre-production engineering studies,
- identifying lower-distortion welding sequences more quickly,
- supporting efforts to reduce the need for physical trial-and-error,
- and enabling a more controlled new-product introduction process.
In other words, the value of artificial intelligence is not only in computational speed.
It is also reflected in more efficient use of engineering time and resources.
Does AI Make the Decision Instead of the Engineer?
No.
One of the key principles behind OSKİM’s approach is the proper positioning of human expertise and artificial intelligence within the same engineering process.
An AI model can compare a large number of results in a short period of time, identify relationships between alternatives, and accelerate the selection of scenarios that should be reviewed in greater detail.
However, determining whether a result is:
- physically meaningful,
- manufacturable,
- compatible with process limitations,
- and applicable under real production conditions
still requires engineering knowledge and shop-floor experience.
The core of the approach is therefore based on the combination of:
Engineering Experience + Simulation + Artificial Intelligence
AI does not replace the engineer.
Instead, it helps the engineer make faster and better-informed decisions.
From Trial-and-Error to Data-Supported Engineering
Manufacturing processes have traditionally been developed through a combination of experience, previous applications, and physical trials.
This accumulated knowledge remains extremely valuable.
Today, however, simulation, data analytics, and artificial intelligence provide new ways to strengthen and extend that engineering expertise in the digital environment.
OSKİM’s welding sequence optimization study is a clear example of this transformation.
Evaluating different welding alternatives before production and analyzing the results with AI support helps enable a shift from trial-and-error-based engineering toward more systematic and data-supported decision-making.
The information generated through these studies can also create value beyond a single project.
Engineering knowledge created and stored digitally can contribute to a growing internal knowledge base that may support future evaluations of similar manufacturing problems.
First Place at WINOVATION 2026
The OSKİM / Dynaress project “Reduction of Distortion in Automotive Control Arms through Simulation- and AI-Assisted Welding Sequence Optimization” was awarded First Place at the WINOVATION 2026 Competition, held as part of WIN EURASIA.
This achievement represents more than recognition of an engineering method.
It also demonstrates the value that can be created when manufacturing experience is combined with digital technologies, simulation, and artificial intelligence.
At OSKİM, We Are Shaping the Future of Manufacturing Today
Competition in the automotive industry is no longer driven solely by manufacturing capacity.
The ability to develop faster, analyze more accurately, anticipate potential problems before production, and strengthen engineering expertise with digital tools is becoming increasingly important.
At OSKİM, we continue to combine our manufacturing experience with engineering simulation, artificial intelligence, data analytics, and digital manufacturing technologies.
The welding sequence optimization project is one concrete example of this approach.
For us, the real value of artificial intelligence in manufacturing does not lie in making decisions independently of engineers.
Its value lies in bringing together reliable data, physical engineering knowledge, and digital technologies to support faster and more informed engineering decisions.
The factories of the future will not be defined only by how much data they generate.
The real difference will be created by those that can learn from their data and turn that knowledge into better manufacturing decisions.
OSKİM Will Be at Automechanika Frankfurt 2026!
At Automechanika Frankfurt, we will showcase our suspension and chassis solutions for the OEM and IAM markets, backed by our strong manufacturing capabilities and extensive experience in the automotive industry.
We invite industry professionals to visit our stand to explore new business opportunities and discuss potential collaborations.
Our Product Groups:
- Control Arms
- Rubber-Metal Components / Bushings
- Ball Joints
- Stabilizer Links
- Shock Absorber Brackets
- Spring Seats
Automechanika Frankfurt 2026:
Date: September 8–12, 2026
Hall / Stand: Hall 12.1 // Stand C64
For more information or to get in touch with us, please visit the OSKİM Contact Page.
Sources
- https://www.nist.gov/publications/2026-roadmap-artificial-intelligence-and-machine-learning-smart-manufacturing
- https://www.nist.gov/news-events/news/2024/05/nist-pursues-ai-enhanced-monitoring-manufacturing-processes
- https://doi.org/10.1007/s10845-022-01963-8
- https://www.nist.gov/programs-projects/augmented-intelligence-manufacturing-systems-aims
- https://www.nist.gov/programs-projects/artificial-intelligence-ai-manufacturing
- https://www.nist.gov/publications/cognitive-work-future-manufacturing-systems-human-centered-ai-joint-work-models
- https://doi.org/10.1016/j.engappai.2022.105142
- https://doi.org/10.1007/s40194-025-02320-y
- https://doi.org/10.4271/2019-01-0818
- https://www.oskim.com.tr/tr/oskim-dynaress-winovation-2026da-birincilik-odulunun-sahibi-oldu/
