New materials are driving technical innovations. Think of high-strength steels with good formability for lighter components in the mobility sector or the use of carbon concrete for the renovation of sewer pipes and in bridge construction. But the development of new materials poses major challenges for materials science: The methods and experiments are costly and it often takes years before a new material is ready for use. In order to be able to research and develop in a more structured way here, there is an increasing reliance on the use of artificial intelligence (AI) methods.
Neue Werkstoffe sind der Treiber für technische Innovationen.
Artificial intelligence is making its way into industry
Artificial intelligence helps collect and analyze the vast amounts of data generated in industry, identify patterns, and derive insights that enable machines and related processes to adapt to new conditions. Whether in automotive or aircraft manufacturing, in Construction or in the field of medical technology - in many industries, the development of materials with new functionalities and improved performance is essential to produce competitive and environmentally friendly products with low resource consumption. After all, the newly developed materials can provide a basis for innovative products.
Innovation through new materials: How industries can gain competitive advantages through AI
In aircraft manufacturing, manufacturers are increasingly turning to carbon-fiber-reinforced composites. They weigh about one-fifth less than steel while exhibiting greater strength. At the same time, research is being conducted on new sustainable bio-materials that are based on renewable raw materials and are biodegradable at the end of their life cycle. Lightweight materials are increasingly being used in architecture and construction, and there is a trend toward personalized medicine in the healthcare industry, to name just a few examples.
Material suppliers and manufacturers are increasingly being asked to address many aspects simultaneously. New materials are supposed to help produce more environmentally friendly and energy-saving materials, they are supposed to produce new functions or be multifunctional in use and have certain properties depending on the application. But it often takes years before a new material can be brought to market maturity. This process needs to be accelerated. The use of artificial intelligence is expected to shorten development times and reduce the costs until new materials can be used. Whether material databases or machine learning - on the way to faster results, various AI methods contribute.
Material databases as a basis for structured work
Continuous sampling and elaborate testing ensure quality in the production process . The challenge here is the often very heterogeneous data that constantly changes during the product life cycle. At the same time, the mechanical properties of materials are determined by their microstructure and thus indirectly by the preceding process or processing steps. This can be remedied by a digital knowledge base. By using material databases, a structured and stringent material development can be realized.
Material data in the databases are only meaningful if they are recorded together with all the process parameters that have acted on the material during production. After all, these general conditions contribute to the structure and thus to the actual properties of the component.
Material data space: basis for materials testing and analysis
In a created material data space, all data can be clearly classified on the basis of a common language rule. By means of knowledge graphs, it is possible to logically link the data. Specialist lights can thus link all accumulating data of the products, processes, machines and sensors along the value-added chain, which is intensive in terms of materials. "Such material databases have a great benefit and enormous potential for identifying previously unrecognized cause-effect relationships between material and product properties in the future," confirms Prof. Dr.-Ing. habil. Marion Merklein, FAU Erlangen-Nuremberg, Speaker Cluster New Materials .
Machine Learning in Materials Analysis and Development
Also in modern materials microscopy, the processing and analysis of data collected in the materials research is one of the greatest challenges. This involves volumes of data in the form of images, samples and components that can no longer be captured by humans alone. The use of machine learning (ML) promises to remedy the situation. This method helps to recognize patterns in data and to derive correlations independently. In this way, evaluations that are time-consuming or repetitive can be automated. In some cases, machine learning even makes analyses possible in the first place.
Such material databases have great utility and enormous potential for identifying previously unrecognized cause-effect relationships between material and product properties in the future.
Superconductors: Energy gain through lossless current conduction
The example of superconducting materials shows how successfully AI can be used in materials research. Thanks to this property, these materials can conduct electricity without loss. However, extremely low temperatures in the range of absolute zero at minus 273 degrees Celsius are required for this. More or less by chance, scientists discovered a few years ago a high-temperature conductor made of copper oxide sandwiches that becomes superconducting at temperatures as low as minus 196 degrees Celsius. Other superconductors were identified that even became superconducting at minus 13 and minus 23 degrees. The only thing that has not been explained so far is what physically happens to make substances exhibit this special property of superconductivity above a certain temperature. Enlightenment is expected from the use of KI . If it is possible to develop materials that already become superconducting at room temperature, huge amounts of electricity for cooling could be saved, since the transition temperature is already below room temperature and the superconductor no longer needs to be cooled. This would be a groundbreaking development for the future.
Artificial Intelligence in Additive Manufacturing: Expanding the spectrum of materials
In additive manufacturing , workpieces are built up layer by layer based on digital 3D design data. This makes it possible to produce highly complex structures that are simultaneously extremely lightweight and stable. But developing materials that are specifically optimized for additive manufacturing requires a very large amount of work. Many additive processes, such as powder bed melting or powder buildup welding, use materials in powder form. It must be ensured that the powder properties are consistent and thus the component properties are reproducible. Handling and understanding the interactions between material and process require further development work to understand and reduce variations in the process and component properties. For this reason, there is increasing reliance on the use of artificial intelligence. Combining artificial intelligence and 3D printing , it is possible to expand the range of compatible materials to that of compatible materials and, for example, control the process based on material response. Thus, the needs of industries such as aerospace, which require safe and reliable materials with the best properties, can be met.
Learn more about materials for tomorrow's innovations:
Artificial Intelligence in Practice: The Possibilities of New Materials and Processes
Whether developing highly heat-resistant materials, chill casting aluminum parts or implementing bionics - the following three use cases show how AI provides very practical support in materials development.
1. Machine Learning in Additive Manufacturing
With futureAM - Next Generation Additive Manufacturing - the Fraunhofer Institute for Material and Beam Technology (IWS) has launched a project to develop very stable and heat-resistant materials. The goal is to replace very expensive materials with less costly ones. With the use of machine learning, it is possible to determine the optimal process parameters, e.g. for the laser powder cladding process.
2. Digital twin replaces costly simulations
In permanent mold casting of aluminum parts, precise component properties are crucial. This usually requires elaborate simulations based on the trial-and-error principle. In a use case coordinated by Fraunhofer IWM, researchers have shown that these loops can be avoided with the help of a digital twin - a digital data model of real material and process data. For this purpose, a data space architecture was used that can be easily transferred to other material processes. The result shows that materials can be integrated into digital value chains in the sense of Industry 4.0.
3. Machine learning gets wings
Optical glasses and the displays of industrial computers have to function perfectly even in humid environments. Promising success here is a nanostructured glass modeled on nature : the self-cleaning and completely transparent wing of the glass wing butterfly. To optimize the individual production steps, machine learning with the help of a computer algorithm was used.
Numerous institutes and companies in Bavaria already benefit from active cooperation in the New Materials Cluster . As a Bavaria-wide information, communication and technology platform in the field of new materials, the transfer of technology across materials and industries as well as proactive networking activities can be advanced together with stakeholders, partners and the cluster advisory board, and current research results can flow into industrial applications. Technology-oriented collaborative projects and regional events also take place at the initiative of the cluster. With the partner package, the New Materials Cluster offers companies and scientific institutions a customized range of services related to innovations in the field of materials.
Accessibility
Language
