May 15 2025

A Glimpse into the Future - Modern Conveyor Systems and Virtual Robotics

 

The AIMS5.0 team at Stralsund University of Applied Sciences is working on the monitoring of very slow-rotating ball bearings as well as complex simulations for mobile robots.

 

Predictive Maintenance for Modern Conveyor Systems

In wafer cassettes or FOUPs, silicon wafers, the "blanks" for microchips and other electronic components, are transported over long distances through the cleanrooms of semiconductor factories. This transport is done using conveyor systems. These conveying systems are extend over kilometers, have a large number of drive units, and almost countless ball bearings. Robustness against failure is essential for conveyor systems, as a failure of a conveyor can quickly lead to a backlog of material flow, thus blocking large areas of production. Methods for predictive maintenance are therefore in high demand to avoid costly downtimes. This is where Use Case 13: "Intelligent Sensors improving Robustness of Automated Wafer Transportation and Storage Systems" comes in.

In this use case, Stralsund University of Applied Sciences is explicitly focusing on roller conveyors. With this type of conveyor system, the drive units only rotate when needed, i.e., when a FOUP passes, and at a very slow rotational speed.

A commonly used approach for detecting damage in rotating ball bearings or rolling element bearings is the detection of vibrations. In many systems, the frequency of these vibrations is within a range audible to humans, so damage detection works with acoustic analyses. The challenge with the investigated conveyors lies in the maximum shaft speed of approximately 120 rpm and less. The vibrations that occur here are low in energy and, in combination with the low frequency and short running times, difficult to detect. The simultaneously occurring ambient noise level is significantly louder at approximately 70 dB(A), dominates the sound signals recorded by the sensors, and further complicates the evaluation.

After many experiments with a wide variety of sensors, the team of Stralsund UAS decided on piezo structure-borne sound microphones. These sensors offer high sensitivity to structure-borne sound and are less susceptible to external airborne noise interference. Just like an instrument maker, we thus determine the smallest change in noise and detect it in comparison to undamaged bearings. To develop a robust diagnostic system, the damage patterns typical for ball bearings were also defined. These include cage damage, inner ring and outer ring damage, damage to the rolling elements, as well as combined forms of damage. The goal is to generate measurement data that enables a clear classification of these damage patterns.

In addition to the further development of the diagnostic system, we are preparing damage to rolling element bearings under controlled conditions to generate and acquire the structure-borne sound data under reproducible conditions. The aim here is to find a clear change in the measurement data that correlates with the progression of bearing damage, from the initial damage to the failure of the bearing.

In parallel, the team is currently collecting data for the construction of a digital twin of our laboratory environments. The plan is to integrate the results data from our roller conveyor monitoring into this twin. In our vision, it will be possible in the future to walk through the production environment carrying an augmented reality headset. With the help of the digital twins, the rolling element bearings with damage symptoms are highlighted and their expected remaining service life displayed.

Furthermore, we want to integrate even more production and mobile systems into this virtual environment. The digital twin of our laboratory environment will also be used to simulate and optimize the interaction of mobile robots with each other and with the environment. This is part of UC14 of the AIMS5.0 joint project: "Speeding-up the Benefit of Automation by Virtual Commissioning through Digital Twins". In short, this complex simulation should help us in the future to optimize the higher-level control systems of the plants in a production line more specifically, as more influencing factors can be taken into account.

We were fortunate to be able to exchange ideas with many experts and interested parties on the topic as the project partner Stralsund University of Applied Sciences of AIMS5.0 at the famous ‘Hannover Messe’ (March 31st - April 4th, 2025) at the stand of the state Mecklenburg-Vorpommern. This always provides new inspiration for finding solutions.

The Upcoming Challenges

 

In addition to a lot of programming work, the following challenges remain for the rest of the project duration:

• We still need to damage and measure many rolling element bearings of our standard type with different damage patterns. Without this, we cannot train the AI.
• We are trying to find out which damage pattern and which sensor signal really indicate an imminent total failure of the bearing. The typical bearing load is not so high that they need to be replaced immediately when the first damage is "audible". Predictive maintenance should also prevent premature replacement.
• We still want to transfer and visualize our AGVs and the conveyor systems into a common digital twin. The aim is to create a holistic logistics system.
• We are trying to find out which damage pattern and which sensor signal really indicate an imminent total failure of the bearing. The typical bearing load is not so high that they need to be replaced immediately when the first damage is "audible". Predictive maintenance should also prevent premature replacement.