The digital twin for thermoplastic composites (digitalTPC) is intended to demonstrate this potential by means of the hybrid injection moulding technology, which is currently establishing itself on the market and is capable of large series production, in which fibre-reinforced thermoplastic composite (TPC) semi-finished products are continuously formed and back-injected. In particular, the complex and heterogeneous microstructure of the composite material itself and its influence by the manufacturing process in the manufacture of semi-finished products and structural components poses an enormous challenge for process control and quality assurance and requires the complete digitalization of the entire production process. At the end of the project, a live demo will be provided that collects and integrates various data from diverse sources and applies data processing and analyses like semantic searching, correlation, semi-product fault detection, fault tracking and visualization.

The project digitalTPC aims at the comprehensive and holistic consideration of all sub-process steps, partly taking place at different locations, from the semi-finished product to the component production. 

With the help of suitable process-integrated and cognitive sensor technology, selected relevant material, process and component characteristics (e.g. fiber orientation, pore content, semi-finished product thickness) are to be measured and recorded by the Fraunhofer Institute IZFP, if possible in local resolution throughout the entire real value-added chain. 

In addition to material know-how, the necessary production plant technology and tools for demonstrator structures are available at the Fraunhofer Institute IMWS (tape semi-finished product production) and ICT (tape laying, consolidation and injection moulding technology) on an industry-relevant scale. The division of the value chain between these two institutes also reflects the industry-typical relationship between material/semi-finished product manufacturers and component manufacturers. 

Fraunhofer SCAI first organizes sensor, machine and simulation data using ontology and then processes the raw measured sensor using AI-based methods and, matches it semantically to a continuous simulation chain across all process stages and provide effective visual feedback to the engineers and technicians of the processes and machines as part of the digital twin.

• Application of MPCCI Mapper in the CAE chain of the project
• AI for fault detection on semi-finished products.
• Data organization through ontologies
• Semantic search engine based on the ontologies
• Automatic Digital Twin for data understanding and comparison

The essential innovation is the comprehensive consideration of the both real and virtual process chains, which enables the individualization of the relevant characteristics of material, process and component. Initially, tolerance limits are used, which multiplicatively link imperfections in every process step along the chain, leading to losses in material and cost efficiency. In this project, the proposed fast digital twin leads to individualized features and an optimized response to variable inputs. Thus, the optimal values do not have to be regulated multiplicatively in each stage but integratively on the basis of the virtual prognosis only at the end of the chain in the component.

• Ontologies
• Concepts for data management using meta-information
• Transfer deep learning & incremental learning for product fault detection

In first February 2022, the first results in the form of a demonstrator will appear.

• Fraunhofer Institute for Algorithms and Scientific Computing
• Fraunhofer Institute for Microstructure of Materials and Systems 
• Fraunhofer Institute for Nondestructive Testing
• Fraunhofer Institute for Chemical Technology
  
   

Advisory Board
 

• Daimler
• PROSTEP
• Dieffenbacher
• Bosch

Project duration: 02/2019 – 01/2022​
Funding Infos: The project was funded by the Fraunhofer Gesellschaft MAVO-Program.