Design

The political and social request for environmentally compatible products is putting companies under increasing pressure. Small and medium-sized companies (SME) in particular have to quickly find or develop solutions to these demands. This paper presents a methodological approach that combines the proven strategies of agile development with the specific activities of so-called ecodesign. The methodology is first discussed theoretically and then experimentally evaluated and discussed by means of a case study in a real company context.

A study by the German Academy of Science and Engineering (acatech) indicates that artificial intelligence (AI) is of growing importance for the success of manufacturing companies [1]. The emerging, data-driven solutions in the manufacturing field are highly diverse, both in terms of the processes and the locations (different factories, factory sub-areas, etc.) where these solutions are implemented. Often the solutions are also hardly scaled beyond the limits defined in the pilot project. When such an AI project ends, the goals of a use case are fulfilled, but this often results in another isolated solution being added to the company’s established IT system landscape. The data this solution delivers is not further used, and complex maintenance requirements negate any gains in efficiency.

In selective laser melting, metal powder is melted layer by layer and fused with the already manufactured part. Within this process, defective layers are created, which can be avoided. Such defects can only be detected by various compression and tensile strength experiments after printing is complete. This procedure is costly and inefficient. Therefore, the authors would like to present a demonstrator which, with the help of machine learning methods which draw from sensor-based data acquisition, is able to detect faulty layers during the manufacturing process itself and to support the machine supervisor with decision recommendations.