Experimental thermodynamic data form the foundation for the development of models in chemical engineering, particularly for data-intensive machine learning approaches, and are essential for chemical process design. Yet these data are scattered across heterogeneous, often unstructured sources, and many chemical engineering tasks, ranging from experimental data handling to process control and decision-making, still require substantial manual effort and are error-prone. Large language models (LLMs) have demonstrated strong performance in knowledge-intensive tasks across science and engineering. In my research, I investigate how LLMs can support and streamline workflows across different areas of chemical engineering.

One focus of my research is the use of LLMs to search for, extract, and curate thermodynamic data from unstructured sources, applying techniques such as schema-guided extraction and retrieval-augmented generation (RAG) within human-in-the-loop workflows. The resulting structured databases provide a basis for process design and the development of data-intensive machine learning models. Another research direction explores the use of LLMs to support process control and management by linking operational data with the documentation and reports that provide their context. Such integrated information can, for example, support anomaly detection in chemical processes. A central challenge across these applications is ensuring the reliability of LLM-generated outputs, which must be verifiable to be trustworthy in engineering contexts.

Lecture Support

• Development and Design of Process Plants (WS 2024/25)
• Thermodynamics II (SS 2025)

Office Hours

By arrangement.

Publications

• L. Vollmer, R. Loubet, F. Jirasek, S. Fellenz, H. Hasse, H. Leitte: Enabling Transparent Problem Solving in Thermodynamics with Ontologies and Knowledge Graphs, Joint Proceedings of the ESWC 2025 Workshops and Tutorials, Portoroz, Slovenia (proceedings).
• R. Loubet, P. Zittlau, L. Vollmer, M. Hoffmann, S. Fellenz, F. Jirasek, H. Leitte, H. Hasse: Using Large Language Models for Solving Textbook-Style Thermodynamic Problems, Comput. Chem. Eng. 204, 109333 (2026). [DOI]
• R. Loubet, P. Zittlau, M. Hoffmann, L. Vollmer, S. Fellenz, H. Leitte, F. Jirasek, L. Lenhard, H. Hasse: Superstudent intelligence in thermodynamics, Phys. Educ. 61, 025004 (2026). [DOI]
• L. Vollmer, R. Loubet, D. Neider, H. Hasse, H. Leitte: Combining Formal Reasoning and LLMs for Scenario-Based Educational STEM Exercises, Hum.-Cent. Intell. Syst. (2026). [DOI]
• L. Vollmer, R. Loubet, H. Hasse, H. Leitte: KnowTD Interface: An Interactive Neuro-Symbolic Workflow for Thermodynamics Problem Solving, Proceedings of the SymGenAI4Sci Workshop at SEMANTiCS-2026, Ghent, Belgium (accepted).

Talks

• L. Vollmer, R. Loubet, P. Zittlau, M. Hoffmann, N. Hayer, S. Fellenz, F. Jirasek, H. Leitte, H. Hasse: Translating Thermodynamic Knowledge to Computers, Thermodynamik-Kolloquium, Stuttgart, 25.-27.09.2024.
• R. Loubet, P. Zittlau, M. Hoffmann, L. Vollmer, S. Fellenz, J. Lenhard, B. Schmid, H. Leitte, F. Jirasek, H. Hasse: Applications of Large Language Models in Thermodynamics, Tag der Verfahrenstechnik, Kaiserslautern, 18.09.2025.
• R. Loubet, P. Zittlau, M. Hoffmann, L. Vollmer, S. Fellenz, J. Lenhard, B. Schmid, H. Leitte, F. Jirasek, H. Hasse: Applications of Large Language Models in Thermodynamics, Thermodynamik-Kolloquium, Hamburg, 22.-24.09.2025.

Posters

• R. Loubet, P. Zittlau, M. Hoffmann, L. Vollmer, S. Fellenz, J. Lenhard, B. Schmid, H. Leitte, F. Jirasek, H. Hasse: Applications of Large Language Models in Thermodynamics, 34thEuropean Symposium on Applied Thermodynamics ESAT 2026, Lisbon, Portugal, 10.-13.05.2026.
• R. Loubet, S. Fellenz, B. Schmid, H. Leitte, F. Jirasek, H. Hasse: Using Large Language Models for Search, Extraction, and Curation of Thermodynamic Data, Thermodynamik-Kolloquium 2026, Dortmund, 30.09.-02.10.2026.
• R. Loubet, N. Hayer, H. Leitte, S. Fellenz, F. Jirasek, H. Hasse: Can ChatGPT Study at RPTU?, 7th Young Researchers Symposium 2026, Kaiserslautern, 15.10.2026.