From Process Analysis at the Edge to KPIs in the Cloud

The More Complex the Process, the Greater the Amount of Data

The analysis becomes significantly more complex and, in particular, more data-intensive in processes with variable behavior. In these processes, the conditions of the different components influence each other. In addition, most production lines have a wide range of different products. “All this makes it very difficult to find reproducible measurement conditions. In these cases, we need methods that use all types of data – for example vibration and process data – and enable reliable and time-synchronous monitoring under all process conditions. The iba system offers a suitable tool for this case, as we have the acquisition and analysis of process and vibration data in one system,” says Reinbrecht. This is the only way to avoid unplanned downtimes and increase plant availability in the long term. The more precisely the data is recorded, the better and more detailed the subsequent process and root cause analysis can be. “The greatest flexibility in analyzing process and machine data is achieved when the data is recorded as high-resolution raw data and not as pre-aggregated data,” explains Reinbrecht. The disadvantage of this recording method, however, is the amount of data it generates. “This can quickly add up to over 100 megabytes per second. For this reason, it is important to extract the information and knowledge from the raw data using efficient methods in order to present the complexity of the process as clearly as possible to different user groups such as production and quality assurance,” reports Reinbrecht.

Hardly Penetrable Interactions: Monitoring a Cross-Rolling Mill

Such complex process behavior can be found in cross-rolling mills: For the production of seamless steel pipes, pipe flaps, i.e. thick-walled hollow bodies, are produced from solid blocks. In this process, a steel block is first heated to a temperature of over 1,200 °C, guided on cylinders, rolled until it cracks centrally and pierced using a mandrel. The mandrel is stabilized and held in position by guide blocks. As all components must function perfectly and in coordination with each other for the process to run correctly, damage or wear to the various units has a major impact on product quality. In addition, further process steps and the necessary cooling time ensure that a lot of time can pass between the occurrence and detection of a deviation – which can have costly consequences in the worst case: If quality control is not carried out until the product has cooled down, several tubes are usually affected and have to be remelted or scrapped. “With such energy and cost-intensive processes, it is important to carry out monitoring in real time. With Edge Analytics, we have the perfect opportunity to calculate the relevant key values from the high-resolution measurement data directly on the process in the edge device, such as an ibaM-DAQ. In this way, we evaluate the data exactly where it is generated and can react quickly and specifically in the event of an error. Robust monitoring with the iba system ensures that, in the best case scenario, we can even intervene before the actual fault occurs by analyzing trends,” says Reinbrecht.

Meaningful KPIs From Raw Values: First the Calculation, Then the Cloud – and Back Again

Comprehensive data analysis is completed by connecting to higher-level IT systems such as cloud environments. This step allows the full potential of the measurement data to be exploited, as the insights gained can be made accessible to different user groups and thus used across the company. However, it should be noted that the information from the large amount of raw data must be reduced to a few reliable indicators: “Due to the immense amount of data, the available bandwidth and the interactions between OT and IT networks, it does not make sense to transfer all the measurement data to the cloud for evaluation, but only the calculated key values. This data can then be displayed in meaningful dashboards using online visualization tools such as ibaDaVIS,” says Reinbrecht. Operators can thus quickly and easily analyze and compare systems, machines and product quality and also identify potential for optimization. However, it is not enough to simply provide the key data. Instead, it is necessary to be able to drill down the KPIs in the cloud back to the raw values. This allows comprehensive root cause analyses to be carried out without loss of information, especially in the event of errors.

Edge Analytics With iba: One System for All Relevant Data

With the iba system, a solution is available that supports the presented method of capturing high-resolution raw data and edge analytics with its architecture and various applications. Based on the structured data acquisition with central time stamping using ibaPDA and the comprehensive process connectivity available here, the raw data can be processed directly in the ibaDAQ or ibaM-DAQ edge devices. ibaPDA offers a powerful expression editor for linking signals and calculating statistical parameters in real time. The ibaInSpectra add-on is also available for online vibration analysis and the ibaInCycle add-on for monitoring cyclic processes, so that all relevant data can be recorded and monitored in one system. Alerts can be sent directly from the edge devices via various output interfaces. Finally, with ibaDaVIS, the iba system has a web-based analysis tool that can be used to create individual dashboards with different graphical elements to evaluate the long-term behavior of the monitored processes and product quality. Interactive and flexible filter options allow the characteristic values calculated in the edge device to be analyzed according to any criteria and technological and temporal filters. With one click, users can jump from the characteristic values back to the raw values and thus enter the in-depth analysis.