Automated Welding Gap-Based Optimization Technology in an Automotive Company: The Wiresense Case and Development of a Welding Parameter Selection Algorithm

The gap between welded components plays a crucial role in determining weld quality, as variations in gap size can lead to issues such as incomplete penetration, incorrect heat input, or distortion. Wiresense is an innovative welding technology that addresses these challenges by using the welding wire as a sensor to measure the gap in real-time. In addition, dynamically adjusting welding parameters, Wiresense ensures optimal weld quality, particularly for thin materials and variable gaps. This thesis, developed during an internship at a leading automotive company, explores the integration of Wiresense into Body-in-White (BIW) and chassis assembly processes. BIW is a critical phase in automotive manufacturing that involves assembling the vehicle’s structural components. Cold Metal Transfer (CMT) welding, known for precise heat control and reduced spatter, was the primary method used. Wiresense enhances the CMT process by dynamically adjusting welding parameters during operation. The thesis details the activities and challenges across three phases of Wiresense implementation. The first phase tested the technology’s initial capabilities in a controlled environment, improving weld quality for previously non-automatable joints. The second phase applied Wiresense in body assembly production lines, integrating it into robotic cells testing the technology ability to improve the welding process. The third phase included tensile and fatigue tests on chassis subframes, demonstrating Wiresense’s potential in improving weld quality for structural components. In parallel, an algorithm was developed to predict key welding parameters based on the gap. This involved, after collecting and refining data from laboratory tests and production samples, to use different statistical method to create the algorithm. While initial results closely matched predicted values, further refinement is needed, particularly for welding speed and heat input. Future steps focus on refining both hardware integration and the algorithm, expanding the dataset, and deploying Wiresense more widely.