Fraunhofer Institute for Production Technology IPTResearch project "GlasAgent"
AI-Driven Development of specialty glass materials
The glass industry faces dual pressure to innovate: reducing emissions while supplying materials for increasingly complex applications. Sensor and medical technology, as well as optical systems for the defense sector, require glass materials whose behavior can be precisely controlled. Developing such materials requires significant time and resources, and traditional methods are often unable to meet the requirements.
The GlasAgent research project is developing an AI-assisted system that supports and accelerates the development of new glass types, optimizing them with respect to ecological and economic criteria. GlasAgent uses simulations and artificial intelligence to determine the optimal glass composition and suitable manufacturing processes. The software agent combines ecological goals with industrial requirements, supporting companies in designing glass in a targeted and resource-efficient manner. To validate GlasAgent, three real glass demonstrators will be manufactured and their properties experimentally verified.
GlasAgent: Digital assistance for glass development
GlasAgent operates through a chatbot interface. Users describe their application, and the system accesses a comprehensive knowledge base. It analyses available data, selects suitable tools and recommends glass recipes, process steps and optimisation pathways. At the same time, GlasAgent evaluates how companies can reliably use recycled raw materials despite their fluctuating compositions. It demonstrates how to close material cycles without compromising functionality.
This is based on a glass ontology developed in the predecessor project, GlasDigital. This ontology brings together knowledge from publications, patents, open data sources, established databases, and real manufacturing processes. In the current project, the team is significantly expanding the ontology to incorporate data on industrial raw materials, analysis and manufacturing processes, simulation-based models, and glass recycling. This creates a robust knowledge base that realistically maps modern glass production and enables reliable recommendations.
Simulation-based glass specification
The scientists of Fraunhofer IPT focus on defining the critical material properties required for emerging markets in advanced optical components and systems such as optoelectronic applications. These application-driven property definitions reflect current and future customer requirements and form the starting point for the subsequent development activities within the project.
Building on this foundation, the Fraunhofer IPT team translates application requirements into quantitative material specifications that cannot be met by existing types of glass. These specifications serve as a reference framework for the project partners developing the simulation tools, data-driven models, and manufacturing methods needed to realize glass materials with the defined property profiles. In this way, IPT’s work ensures that all downstream developments within the consortium are consistently aligned with real industrial and market-driven needs.
The glass development method is validated using three real optical demonstrators. The Fraunhofer IPT team forms the developed glass materials using precision glass molding, followed by comprehensive characterization. Tests under real-world conditions evaluate forming quality, optical performance, thermal stability and chemical resistance of the glass. This demonstrates whether the newly developed glasses reliably meet the previously defined requirements.
To support this process, the team employs simulation-based modelling approaches grounded in precisely characterised reference glasses and integrated optimisation algorithms. The resulting property profiles are systematically integrated into GlasAgent, thereby expanding its capability to generate application-oriented, manufacturable, and industry-relevant material recommendations.
Overall project results and benefits
By the end of the project, not only will there be software solutions, but also physical demonstrators manufactured with the help of GlasAgent. Each demonstrator will be represented as a digital twin and made available to GlasAgent to extend the database. This creates a modular, data-driven ecosystem for glass development, combining industrial requirements with ecological goals to create a new dynamic for the development of sustainable specialty glass.
An overview of some relevant question
GlasAgent enables companies to rapidly select and develop specialized glass materials through an intuitive chat interface. Users can pose application-specific questions such as:
- “Which types of glass remain optically stable at a continuous operating temperature of 250 °C?”
- “Recommend materials for a compact optical system with an Abbe number above 50!”
- “Which forming process is best suited to a lens with low birefringence?”
- “How can I use 30% recycled glass without compromising long-term stability?”
By providing actionable recommendations and guidance, GlasAgent demonstrates how digital and data-driven approaches can streamline material selection and development, reduce experimental effort, and support sustainable manufacturing practices.
Project partners
- Schott AG, Mainz (coordination)
- Fraunhofer Institute for Production Technology IPT, Aachen
- GTT Gesellschaft für Technische Thermochemie und -physik mbH, Herzogenrath
- LTB Lasertechnik Berlin GmbH
- PRALL-TEC GmbH, Lengerich
- Federal Institute for Materials Research and Testing BAM, Berlin
- TU Clausthal, Institute for Non-Metallic Materials TUC
- FSU Jena, Otto Schott Institute for Materials Research (OSIM)
- University of Rostock, Center for Life Science Automation (CELISCA)
Associated partner
Max Planck Institute for Sustainable Materials (MPI-SusMat), Düsseldorf
Funding
The “GlasAgent – Glass Synthesis in the Digital Age” project is funded by the Federal Ministry of Research, Technology, and Space (BMFTR) as part of the “MaterialDigital” initiative.