Industrial Processes and Process Engineering

Industrial Processes are the set of interconnected steps that transform raw materials into finished products. For these processes to be efficient and effective, Process Engineering comes into play. This discipline focuses on the design, optimization, and control of processes, aiming to increase productivity, reduce costs, and improve quality. It acts from the selection of equipment and raw materials to the definition of operational parameters and continuous production monitoring.

Spot Welding and Quality Assurance

A crucial example of an industrial process where Process Engineering and Quality Assurance complement each other is spot welding, also known as resistance spot welding. Widely used in the automotive and home appliance industries, this technique joins metal sheets by applying pressure and electric current, generating heat and forming a weld nugget.

Quality Assurance (QA) is fundamental in this scenario. It encompasses all planned and systematic actions necessary to provide adequate confidence that a product or service will satisfy quality requirements. In spot welding, QA seeks to ensure that each weld nugget meets the required standards of strength and durability.

Quality Defects in Spot Welding and Dynamic Resistance Curve

Even with a well-designed process, spot welding is susceptible to various quality defects. These can be caused by variations in welding parameters (current, time, electrode force), sheet condition (cleanliness, material type), electrode wear, among others. Some common defects include:

  • Metal expulsion (splatter): Occurs when there is excessive current or inadequate pressure, leading to the ejection of molten metal.
  • Cold or insufficient weld: Lack of heat or inadequate pressure, resulting in a weak joint.
  • False adherence: The sheets appear joined, but there is no adequate fusion between them.
  • Excessive nugget enlargement: The diameter of the weld nugget is too large, which can weaken the surrounding sheet.
  • Small weld nugget: Insufficient diameter, compromising mechanical strength.

To monitor and identify these defects, the dynamic resistance curve is a valuable tool. During the welding process, the electrical resistance between the electrodes and the sheets varies. By plotting this variation over time, a dynamic resistance curve is obtained, which provides a unique “behavior” for each weld. Significant changes in this curve can indicate the presence of defects, serving as a real-time indicator of weld quality.

Machine Learning for Quality Classification

Manual interpretation of dynamic resistance curves can be complex and time-consuming. This is where Machine Learning (ML) stands out as a powerful technology for quality classification. Machine Learning algorithms can be trained with a large volume of data, including dynamic resistance curves from good welds and defective welds.

By analyzing patterns and characteristics in the curves, the Machine Learning model can learn to distinguish between a good quality weld and a defective weld. Once trained, the system can automatically classify the quality of new welds in real time, providing instant feedback to the operator. This allows for quick identification of problems, proactive correction of process parameters, and reduction of scrap, optimizing production line efficiency and ensuring that only high-quality products reach the market.