Digital Twin Solutions: As digital images of production systems & production lines, digital twins can make the unused potential in process, maintenance, and quality control visible. They are scalable and can be implemented with low thresholds – making them the ideal technology for production and maintenance.
Digital Twin Solutions: According to a survey by BearingPoint, 75 percent of companies actively deal with the topic of predictive maintenance. This enabled them to reduce their machine and system downtimes by 18 percent and maintenance and service costs by 17 percent.
Digital Twin Solutions: The Solution?
In addition, the companies surveyed noted an average increase in sales of ten percent. The digitization of production can therefore deliver measurable advantages on a large scale. The most demanding tasks in implementing predictive maintenance currently include a suitable IT infrastructure, the selection and availability of data, and the estimated implementation effort. These challenges can be solved through a step-by-step approach and digital twin solutions.
Areas Of Application Of A Digital Twin
The main task of those responsible for production is usually to network production lines or distributed locations and to ensure a seamless production flow and intelligent control of systems and processes – without any loss of quality. Companies use cross-industry digital monitoring in real-time to make adjustments and improvements at any time. Digital twins offer great added value, especially for system and line monitoring, maintenance, and integrative quality control.
In maintenance and repair, the digital twin should help minimize downtimes of machines or entire lines. In addition to on-site maintenance, remote access also has to function smoothly more and more frequently. In addition, predictive maintenance is increasingly being used to predict faults before they lead to problems. A typical solution is an intelligent warning system that triggers an alarm, for example, if machines or sensors deviate from the learned standard behavior. A prerequisite for a functioning warning system is an extensive “data lake” with valid data, which is fed from the production data collected.
In production planning and control, on the other hand, the challenge is to ensure the production lines and feeds on the shop floor, to optimize set-up times, and to carry out start-up simulations with digital support. The basis for this is data from ongoing production, which is collected by connecting machines or retrofitting the systems with sensors. They are then consolidated and processed so that the production process can be presented transparently and optimized based on the data.
Use Of Digital Twin Solutions In Semiconductor Production
One company already successfully using predictive maintenance is the American semiconductor manufacturer Global foundries (GF). He recently started using smart sensors to monitor the ultrapure water valves at the Dresden site. The valves are critical to production and have been monitored by employees at great expense. Global foundries now collects audio data from the valves and then uses machine learning methods to create a data model. This enables – in combination with continuously recorded sensor data – an evaluation of the actual state and a prognosis concerning expected changes of the valves.
Experts can view this information via a dashboard and carry out appropriate maintenance measures. In addition, historical and current parameters can be displayed, and limit values for an alarm can be defined. This enables early error detection and needs-based maintenance planning, thereby preventing downtimes, reducing maintenance costs, and significantly increasing the reliability and service life of the system.
When companies decide on such an IoT system, they should note that there is currently no ready-made system that directly meets all the requirements of such a smart factory project. IoT systems in production always have to be adapted. To ensure a quick project start, it makes sense to encapsulate typical functions in prefabricated modules and assets and separate them from customer and system-specific aspects. Examples of such modules can be anomaly detection, AR / VR-based maintenance scenarios, or blockchain-based manufacture proof. Despite modularization and individualization, solutions must essentially be designed flexibly and implemented so that they are scalable. That means: Entry must be inexpensive with just a few systems and quickly show added value.
The Way To The Digital Twin
Therefore, the development of seamless monitoring of production systems corresponds more to a step-by-step process. In the beginning, there is the recording and consolidation of data from every single machine, every single sensor, and so on. The second step transfers them for processing on-premise via an edge controller in the periphery or directly to the cloud. In the third stage, all structured and unstructured data are processed, visualized, and interpreted to make them usable for process optimization. Experts can now recognize patterns, establish correlations and determine threshold values for alarms. In the fifth and final phases, a fully digital image of the existing production processes can be “learned” using AI and machine learning.
The status and behavior of production systems can be monitored on the digital twin and intelligently analyzed and forecast thanks to AI or machine learning. Companies can thus minimize physical, retrospective quality controls, make maintenance processes more efficient and keep an eye on production facilities across all locations. The entry barriers to setting up a digital twin are very low, thanks to the scalable step model and the immediate added value.