Customer application 16 December 2022
Looking to the Future

WAGO has substantially reduced its maintenance needs for housing production using a predictive analytics system developed in-house. The system also provides valuable services in early fault detection and process optimization. WAGO also offers its analytics solutions to other companies.

WAGO products are used globally in factories, ships, rail and electrical networks, from the desert to the tundra. Sometimes, they endure a lot: dust, dirt and shocks, as well as scorching sun, icy winds and the muggy tropics. Protection against the harsh environment comes from tailor-made housings, which WAGO – like many other plastic components – often manufactures itself.

Various plastic granules are the base material for the housings; these granules are transported to the injection molding machines by a tube system. This occurs via vacuum pumps, which produce a vacuum. Like the good old pneumatic tubes, the pumps transport the different granulates to where they are needed immediately. However, the airflow inevitably also carries some dust. To prevent this from damaging the pumps, filters are installed in the exhaust air pipes.

These filters must be carefully cleaned repeatedly; otherwise, the conveying capacity suffers “This is like a vacuum cleaner: the fuller the filter, the worse its performance and efficiency,” explains Sebastian Pscheidt, Technical Engineer for injection molding technology at WAGO. In the worst-case scenario, the tube is not completely emptied when another granulate is conveyed, mixing the two materials. WAGO staff have so far prevented this by cleaning the filters at fixed intervals. Often, however, this would not have been necessary because the filters still allow sufficient air to pass through – annoying because the cleaning is labor- and cost-intensive. In addition, the material distribution system must be shut down during this time, which can disrupt housing production.

MM-339227_new_website_page_WAGO_1S7A5285_2000x1125_.jpg

Our predictive maintenance solution ensures that we can now intensively clean the filters as required. With such predictive planning, we significantly reduce maintenance costs – and, at the same time, increase process reliability.

Dr. Jan Jenke | Product and Project Manager Analytics at WAGO

“As Much Effort as Necessary, as Little as Possible”

Good reasons for WAGO to develop a condition monitoring system – using WAGO technology, of course – to enable predictive maintenance. “As much effort as necessary, as little as possible,” is the strategy as Dr. Jan Jenke, Product and Project Manager Analytics at WAGO sees it. The filters are only cleaned when it is expected that the output will fall below an acceptable level. For this, WAGO primarily uses data from sensors that measure the pressure upstream and downstream of the filter. Sophisticated analytics methods can then derive forecasts on pollution trends. The system then automatically triggers a maintenance order in the SAP system for just-in-time maintenance. This ensures that the cleaning always occurs at the optimum time. Our predictive maintenance solution ensures that we can now intensively clean the filters as required.“Our predictive maintenance solution ensures that we thoroughly clean the filters as required. With such forward-looking planning, we reduce maintenance costs considerably while increasing process reliability,” Jenke concludes. A bonus is that the system also saves energy because the filters are cleaned earlier when more dirt is present, so that the pumps need less power. It is not possible to exactly project how high the savings will be because this depends on many factors, explains Pscheidt. “But since our pumps often work with several kilowatts of power, the savings are quite significant.”

Fault Detection with Machine Learning

WAGO also uses the data from these and other sensors in the material distribution system for additional purposes - including the early detection of possible faults during conveying. Previously, the employees worked with Excel lists. Machine learning models now handle this task: They automatically identify anomalies in the data before major disruptions occur. This reduces the time required for troubleshooting by more than fifty percent. “While predictive maintenance is about looking into the future, fault detection with machine learning is about detecting past and present data patterns,” explains Jenke. The system is designed to be open in such a way that additional machine learning models can be added if required. A dashboard displays the data and provides an instant overview of the process quality. The system automatically flags detected anomalies. However, the dashboard offers numerous other visualizations to maximize the information's benefits. This enables employees to perform individual evaluations and modeling via a user-friendly interface with live and historical data from any time period. The result is greater transparency, for example, regarding the system's capacity utilization. Above all, it is easy to see where there is potential for optimization. “With the two analytics solutions and the dashboard, we have provided employees with a digital toolbox that simplifies their work significantly. They will gain a better understanding of cause-and-effect relationships of their actions,” says Pscheidt. At the same time, the dashboard promotes cooperation between WAGO's process, automation and analytics experts.

“With the analytics solutions and the dashboard, we have provided employees with a digital toolbox that simplifies their work significantly. They will gain a better understanding of cause-and-effect relationships of their actions.”

Sebastian Pscheidt | Technical Engineer Injection Molding Technology at WAGO

Infrastructure Flexibility

Data acquisition and exchange are performed via WAGO's 750 Series I/O System with several modules, including digital input and output modules, analog outputs, power measurement modules and rail-mount terminal blocks. “The plug-in connection design is very easy to use and ensures unparalleled system uptime,” explains Pscheidt. The data is then analyzed on a locally installed WAGO Edge Computer. Alternatively, it would also be possible to do this in a cloud, on an existing IT infrastructure – or directly on the WAGO Controller. Its Linux^® operating system allows setting up individual Docker containers that bring analytics functionalities to the product.

Analytics Also for Customers

However, WAGO does not only develop analytics systems of this kind for its self – the company also makes its know-how and technology available to others. For example, WAGO provides its customers with individual analysis solutions, which can be used to examine processes, identify weak points, or recognize starting points for optimization. Companies in various industries can use it, for example, to determine the reasons for unusual operating states of their systems and machines, identify impacts on product quality and process stability, or improve machine parameterization. “With our analytical expertise, we help our customers bring more transparency to their processes,” says Jenke. “This will strengthen their competitiveness sustainably!”

Text: Ralph Diermann – Energy Journalist

Photos: Tremonia