Towards a Digital Twin for Individualized Manufacturing of Welded Aerospace Structures

Licentiatavhandling, 2023

The aerospace industry is constantly striving towards lower fuel consumption while maintaining a high standard with regards to safety and reliability. These increasing demands require the development of new methods and strategies for efficient and precise manufacturing processes. One way of achieving this goal is fabrication, an approach where components are built by joining multiple small parts into an assembly. This brings many advantages such as more flexibility in product design, however it also adds geometrical variation to the manufacturing process which needs to be managed. Since the parts in the assembly are produced separate from each other before being joined together, issues can occur related to how these parts fit together. If a single part in the assembly deviates slightly from its intended shape, this deviation may propagate in the assembly. It may also stack with deviations in other parts. This can sometimes be difficult to predict and manage using existing manufacturing tools developed within the fields of geometry assurance and robust design.
The traditional approach to managing geometrical variation is usually based on making statistical assumptions about the variation that is going to occur in the manufacturing chain. With rising complexity in product design and increasingly tight tolerances, the traditional geometry assurance approach may not be sufficient to guarantee the high geometrical quality required from the final product. Individualized manufacturing has previously been proposed as a way of increasing the precision and reliability of a production process by treating each product individually based on its unique properties. This can be achieved with a digital twin, an emerging technology which works by creating a virtual copy of a physical process. The work presented in this thesis is directed towards realizing a digital twin for fabricated aerospace components. The first contribution is a framework describing how a digital twin could be implemented into a typical fabrication process within the aerospace industry. Since fabrication makes heavy use of welding to join multiple parts, welding simulation is an important component in this implementation. The digital twin also needs to manage measurement data collected from the parts on the assembly line, and this data should be considered within the welding simulation. The result of this simulation is then used to adapt and adjust the manufacturing process according to the conditions that have been measured and analyzed. An analysis loop is proposed in this thesis for realizing the functionality of the digital twin. A case study is conducted to evaluate the precision of the proposed analysis loop by comparing its predictions to a real welded assembly. The results of the case study show that the predictive precision of the proposed method beats the accuracy of a traditional, nominal prediction. This is an important first step towards the completion and future implementation of a digital twin for welded assemblies.