Big Data - Big Chance

The volume of machine data generated in production plants is increasing continuously. With its Big Data Project, Hauni is now analyzing this data and generating services that support customers in their manufacturing operations.

Objectives of big data analytics. The vision is to create highly automated, self-learning machines. Hauni is already developing specific applications that combine big data analysis with the company’s unique knowledge of the industry and its machinery.

A PROTOS-M5 generates up to 200,000 data points per second – most of this information is required by the deepest levels of the machine controls and overwritten by the program after a single use. Although these data can be made available for other purposes, until now nobody had considered doing so.

Ralf Heikens, Head of Innovation Center Automation Technology, recognized that it was time for a change. "By linking our know-how with big data analyses of machinery in different locations around the world, we can help our customers to use their machines much more efficiently," he explains. "This allows them to increase output, efficiency and quality, as well as reduce material waste and resource consumption." In order to produce measurable results as quickly as possible, Hauni has entered into a development partnership with the Fraunhofer Big Data Alliance and established a new big data / data science team. This group works efficiently with machine software developers and process experts, deploying an agile approach based on use cases. In addition to customer-specific applications, Hauni has focused on three principal areas and developed concrete solutions for them: anomaly detection, parameter optimization and stop analytics.


Anomaly detection


Hauni’s anomaly detection system is designed to improve the production process in terms of efficiency, production waste and quality. It detects anomalies in the machine data, reports them and provides additional information as a recommended action appropriate for the target group. Since these analyses and statements are based on a broad range of data and experience, rather than the professional evaluation of a single person in the production process, they bring together almost the full spectrum of experience available in various production departments and combine it with Hauni’s know-how. The result is the optimum solution based on up-to-the-minute industry data and expertise. This saves money by eliminating the need for complex and time-consuming error analysis by the individual company.


Parameter optimization


In day-to-day production, parameter settings for individual makers can vary widely – even for identical or similar products and machines – because they are often adjusted manually based on the individual operator’s experience. Taking this into consideration, Hauni’s parameter optimization system focuses equally on increasing machine efficiency and product quality, as well as reducing maker stops and waste. It uses production experience accumulated over many years to find and recommend the most successful parameter settings for similar machines and brands.


Stop analytics


Hauni’s Big Data Project starts by researching the causes of machine stops. "Our assistance system helps to identify the root causes of problems by analyzing stop-related data from a very large database rather basing a diagnosis on personal experience. It recommends measures that ensure shorter repair times and prevents downtimes from occurring again," says Marc Stahl, Project Manager Big Data Analytics.When a machine stops, it displays a process malfunction as the cause. For example, this could be a rod break in a cigarette maker and therefore have about 30 possible causes. The new Hauni system aims to alert the operator to the specific cause, thus enabling them to identify the correct remedy. This saves time because the machine is not started repeatedly with the same error. "As a result, we are increasing the mean time between failure. But we are also reducing the mean time to repair and the number of rejects because several hundred cigarettes are rejected as start and stop waste after each restart," explains Stahl. "This combination of analysis and expertise shows why, as the world’s leading supplier of solutions to the international tobacco industry, we have entered the field of big data analytics – an area that is otherwise primarily the domain of software companies. Software giants have endless capacity and expertise in artificial intelligence design – but they don’t have our technical insights into tobacco machines."