In the business world, there are various reasons for new product requirements: technological innovations, legal changes, or entrepreneurial decisions, to name just a few. Typically, such changes lead to the redesign of products and production processes. Especially in rapidly changing requirements, this time-consuming approach often proves to be ineffective because the planning essentially starts from scratch. Information and data about previous product variants and production processes cannot be fully utilized, partly due to the lack of technical infrastructure.

On the other hand, nature always builds upon what already exists, meaning it relies on existing "datasets" and undergoes evolutionary changes based on them. It also utilizes all adaptations, whether successful or not, to gain insights that it incorporates into further developments.

Natural evolution as a model for flexible, adaptive production systems

In the four-year Fraunhofer flagship project "EVOLOPRO - Evolutionary Self-Adaptation of Complex Production Processes and Products," which was recently completed successfully, over 50 researchers from seven Fraunhofer Institutes analyzed various mechanisms of natural evolution in organisms under changing environmental conditions and applied them to modern manufacturing processes. In addition to Charles Darwin's general theory of evolution, they paid special attention to the "Theory of Facilitated Variation." This theory divides the ability for rapid adaptation into various "elements of flexibility," including modularity and hierarchy.

The researchers utilized evolutionary biological elements and mechanisms to design a new generation of "Biological Manufacturing Systems" (BMS). Biological Manufacturing Systems are capable of autonomously adapting to new requirements and environmental conditions, similar to biological organisms. However, unlike nature, they do not require several centuries to do so. Thanks to the recent advancements of Industry 4.0, the team suggests that adjustments can be made within a short period of time. "The vast field of digitalization creates excellent conditions for the desired evolution in production technology," says project leader Dr.-Ing. Tim Grunwald from the Fraunhofer Institute for Production Technology IPT in Aachen.

Biologically inspired algorithms and the advancement of concepts such as "Digital Twin" and "Digital Environment" as drivers for production technological evolution

To implement Biological Manufacturing Systems, the project partners relied on two main components: biologically inspired algorithms and data-driven Digital Twins that interact with a Digital Environment.

The Digital Twin is the digital representation of an individual component and enables the digital processing of all component-related information. The Digital Environment replicates the key requirements imposed on the component by the real environment. The project team developed a concept for a multi-level Digital Twin, which, in a modified form, was also applied to a multi-level Digital Environment. Both concepts can be flexibly integrated into existing software environments, such as CAM systems.

The biologically inspired "Facilitated Variation algorithms" are mathematical tools, some of which already existed, while others were newly created within the scope of the EVOLOPRO project. These algorithms enable the Digital Twin to undergo an evolutionary path. Operating in the background on a "digital auxiliary level," they run continuously in parallel with the real production process, facilitating a continuous, data-driven learning process.

Validation in three product and process chains

The effectiveness of the concepts and algorithms was tested in three product and process chains: the pilot chain "Aviation," the pilot chain "Optics," and the pilot chain "Automotive." All three test series concluded with successes and new insights.

In the Aviation pilot chain, the project team was able to develop a simulation environment for model-based process planning and design based on the biologized algorithms and concepts. Using this testbench, the team significantly reduced the effort required for process planning and the ramp-up process for milling a Blade Integrated Disk (Blisk), a highly complex turbine component, as well as for variations in the Blisk design.

In the Optics pilot chain, significant improvements were made in the digitization of manufacturing complex glass optics. The team was able to directly incorporate simulated results into optical design and assembly. By bringing planning and implementation closer together, the team gradually increased the flexibility of the overall process. Furthermore, the researchers developed a self-learning method for the automated assembly of optical components that requires fewer steps than any previous assembly algorithms.

In the Automotive pilot chain, a fully model-based controlled bodywork production was established, leveraging the full potential of an automated, self-learning Industry 4.0 body construction application. "The pilot chains had fundamentally different requirements and characteristics. The achieved results thus speak to the universality of the project approach pursued," expresses project leader Tim Grunwald with satisfaction.

Data Lake architecture for cross-site data processing

To centrally store the large amount of collected process data from the three EVOLOPRO pilot chains and facilitate data exchange between Fraunhofer Institutes, the project team established a "data lake architecture." This is a cloud-based application that includes standardized data interfaces and domain-specific description models, known as ontologies, for the unambiguous assignment of uploaded data.

The new cloud architecture enabled cross-site automated data exchange for the teams. Project leader Tim Grunwald sees significant potential in this approach: "In production, uniquely labeled data is often scarce and expensive to create. Additionally, it is very time-consuming to establish a database based on simulations. This is precisely where our technological developments from the EVOLOPRO project can directly contribute to economic value," says Tim Grunwald.

In further research projects building on EVOLOPRO, the concepts of the digital twin and the digital environment will be further developed. The measures developed from the pilot chains will be purposefully advanced by the research teams towards market readiness and industrialization.

Funding

The four-year Fraunhofer lead project "EVOLOPRO - Evolutionary Self-Adaptation of Complex Products and Production Processes" was initiated by the decision of the Fraunhofer Society's executive board in 2018 and was funded from January 1, 2019, to March 31, 2023.

Contributing Institutes

• Fraunhofer Institute for Production Technology IPT (Coordinator)
• Fraunhofer Institute for Mechanics of Materials IWM
• Fraunhofer Institute for Applied Optics and Precision Engineering IOF
• Fraunhofer Institute for Material and Beam Technology IWS
• Fraunhofer Institute for Machine Tools and Forming Technology IWU
• Fraunhofer Institute for Applied Information Technology FIT
• Fraunhofer Institute for Algorithms and Scientific Computing SCAI