One of the main tasks of production managers is to guarantee a seamless production flow and smart control of systems and processes while maintaining a consistently high level of quality and efficiency. Companies ensure this across all industries through digital real-time monitoring. Digital images of processes and devices in production environments, so-called digital twins, can offer great added value, particularly in the application areas of plant and line monitoring, maintenance and integrative quality control.

Optimization in production and maintenance

In production planning and control, the challenge is to secure the production lines and infeeds on the store floor, optimize set-up times and carry out start-up simulations with digital support wherever possible. The basis for this is knowledge in the form of data from ongoing production. This data can be collected by connecting machines or retrofitting the systems with sensors, known as retrofitting. It must then be consolidated and processed into a valid data lake in order to present the production process transparently and optimize it on this basis.

In maintenance and servicing, the digital twin is intended to reduce downtimes of machines or entire lines. In the meantime, not only local maintenance but also remote access must be ensured. In addition, the industry is increasingly relying on predictive maintenance. According to a BearingPoint survey conducted at the beginning of 2021, 75% of companies across all industries are actively addressing this issue. Predictive maintenance is designed to detect potential faults before they lead to problems. A typical solution for this is an intelligent warning system based on the collected production data, which triggers an alarm, for example, if machines or sensors deviate from the learned standard behavior. Companies benefit from flexible, needs-based maintenance instead of rigid service models.

Learninga digital image of production processeswith AI and ML

One company that is already successfully using predictive maintenance is the American semiconductor manufacturer Globalfoundries. It recently began using smart sensor technology at its Dresden site to monitor its ultrapure water valves. The valves are critical to production and were previously monitored analogously by employees at great expense. Globalfoundries is now recording audio data at the valves in order to create a data model using machine learning methods. In combination with continuously recorded sensor data, this enables an assessment of the current status of the valves, known as the "health condition", as well as a forecast of expected changes in status. Experts can view this information via a dashboard and initiate appropriate maintenance measures. In addition, detailed curves of historical and current parameters can be displayed and limit values for alarms can be defined. This enables early fault detection and needs-based maintenance planning, thereby preventing downtime, reducing maintenance costs and significantly increasing the reliability and service life of the system.

Customization of IoT systems

What companies should know: There is no ready-made system that can meet all the requirements of such a smart factory project from a standing start. IoT systems in production always require adaptation, as every production line differs in terms of the machinery used and the associated control systems, data and protocols. Requirements for dashboard visualization are usually just as different as the IT and OT systems and business processes to be integrated. To ensure a quick project start, it makes sense to encapsulate typical functions in prefabricated modules and assets and separate them from the customer- and system-specific aspects. Examples of such modules include anomaly detection, AR/VR-supported maintenance scenarios or blockchain-based proof of manufacture.