Doppelmayr Seilbahnen

Johannes Moritzhuber

The visual inspection of fully assembled components is gaining increasing relevance in contemporary industry. This fact has long been recognized in mass production or safety-oriented industries such as aerospace manufacturing. Monitoring production processes has been a common practice even before the emergence of the term "Industry 4.0." However, applying these methods to custom-made products, often produced in small quantities, would make the associated costs non-competitive.

Undesirable deviations in product quality can quickly become public knowledge and subject to sanctions in today's era of social media. In addition to potential warranty claims, which can reduce profit margins, there are other factors to consider. The reputation of a high-quality product or brand can be quickly damaged. From a process perspective, we need to consider the risks that should be avoided through careful process design. In recent years, the effort to minimize risks has increased significantly. In the past, in custom manufacturing, it was sufficient for experienced employees to assemble the product based on their expertise, with a quality expert inspecting the finished product for quality at the end of the process. Nowadays, process steps need to be verified and checked during their execution since it's no longer possible to verify the correct process parameters at the end. For example, lubricated bearing points with limited access at the end of the process cannot be directly inspected. If an employee does not execute this step correctly, it affects the quality of the finished product. The question is how significant this error ultimately is for the product. If this error is not discovered throughout the product's entire lifecycle, it may have no impact on the product's operation, or it may lead to premature product failure. However, from a manufacturing cost perspective, the situation is different. A small error in a low-cost process step can result in significant follow-up costs, including negative publicity on social media.

To meet these increased requirements outside the established mass production industries such as the automotive industry, we rely on the possibilities offered by digitization. In our analysis, Industry 4.0 should not be just a buzzword or an inflated and costly process. Instead, we have thoroughly examined how we can effectively utilize digitalization methods to establish a new quality standard in the process without increasing the effort in the manufacturing process.

“In our analysis, Industry 4.0 should not be just a buzzword or an inflated and costly process.”

The foundation for achieving this lies in a modular consideration of the process step. When examining an assembly process from a modular perspective, there are usually a limited number of different process steps that are executed with various parameters. An excellent example of this is tightening screws with different torque values. In process analysis, these are referred to as process parameters. Besides torque, there are other parameters involved in screw tightening. However, I don't just want to focus on the technical aspect but also on the definition of the process itself. If we consider the proven traditional approach in terms of Industry 3.0, in this case, the process would typically be defined by a work preparation specialist. This usually only occurs for particularly critical steps and not in the overall context. The effort and, in particular, the required expertise would exceed the time frame available in day-to-day operations. In reality, we rely on the knowledge of the employees. Regardless of the fact that we can no longer find enough skilled workers with extensive experience today, the question arises whether a skilled worker can even assess the risk associated with this process step. In our view, this is not the case, especially because it requires not only expertise but also comprehensive knowledge of custom product variations and increased requirements from standards.

Therefore, we first isolate each individual step and examine the various parameters and variations present in the execution of that step. We employ a process FMEA and involve many knowledge carriers within the company to design a decision tree that suits our product. The result of this analysis is the definition of a process module with a decision architecture comparable to a decision tree. With the help of this decision tree, when the process step is called up in our digital assistance system on the shop floor, the step is controlled according to the requirements of the entire process, and the employee is given appropriate instructions based on the step's demands. The requirement is defined by product characteristics or parameters from the PLM (Digital Twin) or ERP system. An MES system reads the product characteristics from the systems, and the parameters of those characteristics then control the decision tree. As a result, a digital instruction for the employee and the tool is generated.

Let me illustrate the advantage of this approach with a simple example using screw tightening. When the process module is called up, the selection of the tool used, the screw tightening parameters in the tool, and additional inspections at the end of the process are defined based on the screw tightening requirements. Additionally, the employee may be instructed on the screen to mark the screw connection with a colored stripe. Furthermore, other subsequent process steps may be assigned tasks to complete the process step. In this example, a downstream robot equipped with a vision smart sensor would verify the colored marking of the screw connection. Simplified, this process would already be the result, but where does the detailed advantage lie? In this example, achieving the correct torque at the end of the process should be sufficient. However, the path to achieving that goal can have different requirements. An experienced employee knows from their experience that this particular screw connection is problematic in execution. In the past, there have been issues with this specific screw connection. Therefore, the employee uses a better tool to tighten the screw and double-checks the torque at the end. To ensure that they have properly tightened and checked the screw, they mark the connection with a colored stripe. This way, they or a quality assurance employee can quickly verify if this particular screw connection was tightened perfectly.

The process module specifies this approach for screw tightening with higher requirements. Additionally, many parameters are collected as a byproduct of digitization for documentation purposes. This further enhances the value of the screw tightening process. If the screw connection does not require this level of effort, the instructions for the employee and the process requirements would be much simpler. In summary, through automated process design enabled by the digital product characteristics of the Digital Twin, processes can be digitally tailored to their value. This not only increases quality by aligning the execution of the process step precisely with the requirements but also enhances manufacturing efficiency. Only the process execution necessary for the requirements is selected. Additionally, any skilled worker, even without experience, can perform these work steps without the need for specialized departments to create an individual process flow in the background.