Cookie
>Home

GlobalFoundries Dresden

Not to run out of Chips:
Acoustic Diagnosis While driving by

Chips are everywhere – in household appliances, smartphones, cars, industrial plants and medical devices – and highly requested on the global market. In the largest and most modern semiconductor plant in Europe, GlobalFoundries manufactures microchips for customers in these and other fields. In order to optimize maintenance cycles and prevent production outages, T-Systems MMS has developed a completely new type of cloud-based IoT platform together with the Smart Systems Hub. Conceived in pioneering work within the framework of the proven “Digital Product Factory” co-innovation format within three months, the smart AI solution enables the highly automated driverless transport system in the cleanroom to be maintained and serviced proactively.


Customer Benefits


Checkmark for the benefit of our customers

Reduced maintenance and servicing effort due to optimized, demand-oriented maintenance cycles

Checkmark for the benefit of our customers

Early detection of maintenance requirements increases reliability, avoids production interruptions and increases plant utilization

Checkmark for the benefit of our customers

Positioning as one of the leading semiconductor producers within the meaning of “Smart Factory”

  • Lars Fienhold | Principal Analyst Factory Automation | GlobalFoundries Dresden

    „Smart maintenance is on everyone’s lips, only you have to do it. I am absolutely thrilled to see how this project was able to deliver results within the short timeframe for our very special transport system. The business case is great: the solution can reduce downtime in the affected areas by 25 to 35 percent!“


Reference at a glance


Task

Task

Reduction of maintenance and servicing costs for the wafer transport system in the cleanroom

 

Solution

Solution

Build an AI-supported cloud infrastructure for real-time acoustic monitoring and anomalies detection

 

Benefits

Result

Permanent monitoring of the health condition enables demand-oriented, predictive maintenance and improves plant utilization

 

  • Requirements

    Reduction of Maintenance Effort

    The manufacture of microchips involves more than 1,000 individual steps under the strictest cleanroom conditions. The complexity of the entire production process, in which interdependencies exist in sequence, is correspondingly high.
    The 300mm wafers, from which the microchips are made, are transported in overhead transportation vehicles during the approximately 1,400-step manufacturing process. On a 22-kilometre rail network on the roof, the “Automated Material Handling System” (AMHS), more than 800 vehicles transport the wafers in special boxes every day. At each machining station, a lifting system lowers the boxes. The wafer boxes are then automatically transported to the next production step. If a robot fails during operation, it would block parts of the rail system in the clean room and thus slow down production. Up to now, the complex mechanics of the vehicles have been monitored manually and maintained in a precautionary manner.
    To take another step towards the Smart Factory, GlobalFoundries commissioned T-Systems MMS and the Smart Systems Hub to develop an intelligent IoT solution. In order to be able to maintain the wafer transport system in the future and to optimize maintenance cycles, the transport system should be monitored live and its condition visualized in real time.

  • Solution

    Acoustic Diagnosis in Driving by

    The “GlobalFoundries Health Predictor”, developed jointly by T-Systems MMS and the Smart Systems Hub, was installed on a rail of the wafer transport system in the clean room during operation. While the Smart Systems Hub supplied the hardware, T-Systems MMS supplied the software and AI components.
    With the smart solution, a highly sensitive acoustic edge computing sensor detects the sound of passing vehicles and can detect anomalies, such as when a wheel is damaged or a ball bearing is defective. Using machine learning and artificial intelligence, the vibration and vibration data are compared with the historical data already available in a database. In contrast to a cloud solution, edge computing helps to keep data transmission latency low enough to detect and analyze errors in real time. On a dashboard, the “health status” of the transport robot is permanently monitored with a traffic light system. The cloud platform Amazon Web Services AWS is used for this purpose. Various data sources have been integrated into the AI-based cloud solution, including innovative MEMS acoustic sensors and a big data application that evaluates more than 10,000 vehicle-related data files per day.

Expected failures and maintenance requirements can now be predicted by triggering an alarm when approaching defined threshold values and identifying transport robots with a high probability of failure. The error data collected is also used for reporting purposes, such as analyses, reports and audits.
The pilot solution “GlobalFoundries Health Predictor” installed at the Dresden semiconductor plant first served to clarify the technical feasibility. In the meantime, a follow-up project is being implemented in which edge computing sensors are being installed in several places, for example in curves and in various rail systems. If successful, the rollout of the IoT application is also planned for other GlobalFoundries production sites around the world, such as New York and Singapore.




  • Benefits

    Maximum Output due to Smart Maintenance

    Without having to intervene in ongoing production, the maintenance needs of the transport robots are identified early, precisely determined and visually displayed. Long before damage occurs, the Health Predictor registers abnormalities in real time. Maintenance and servicing of the highly complex wafer transport system can thus be aligned to actual needs, maintenance cycles can be optimized and the availability of the transport system can be further maximized.
    Thanks to the Health Predictor, chip production can be further optimized and production capacities can be utilized to the maximum – especially given the current shortage of chips on the world market. In addition, resources are reduced in terms of material, personnel and time. This increases the lifespan of transport robots as a result of timely maintenance, and intact parts no longer have to be routinely removed. A fact that also takes sustainability into account. At the same time, the on-site presence of maintenance engineers in the cleanroom can be significantly reduced with the highest manufacturing quality. As the hardware and software components of the Health Predictor are modular, the AI-based application can also be used for other manufacturing and logistics systems of The semiconductor plant in Dresden can position itself as a pioneer in the field of smart factory and sensor-based monitoring both within the company and in the industry as a whole.

About GlobalFoundries

GlobalFoundries is an American semiconductor producer based in Dresden, Germany. The company employs more than 15,000 people worldwide. In addition to the plant in Dresden, there are three other production sites in the USA and one in Singapore. More than 3,200 technicians, engineers and specialists from all over the world work on the company campus in Dresden. GlobalFoundries has helped establish the Free State of Saxony as the leading micro- and nanoelectronics center in Europe and has significantly advanced the development of new IoT and connectivity system solutions with the Smart Systems Hub.


Predictive Maintenance at GlobalFoundries
Predictive maintenance of
ultra-pure water valves
Link to AI-based solution

Contact Person