Energy transition, raw material conversion, short product cycles – the chemical industry needs new and sustainable processes quickly. Downstream processing is often a complex and energy-intensive bottleneck.
In the DiKey research project, the consortium, including INOSIM, uses material data and machine learning to identify the most promising separation processes, optimizes them through modeling, and tests their implementation for process development.
The joint project DiKey – Digital Key Technologies for Determining Material Data for Efficient Material Separation in the Chemical Industry started on April 1, 2025, and will last until the end of March, 2028.
Predicting material data for the industrial transformation
In times of energy transition, new raw material sources, and increasingly shorter product life cycles, the DiKey project aims to develop new solutions based on material data. These are of fundamental importance for the chemical industry, as they determine which new processes are even feasible. However, given the large number of possible mixtures, it is not realistic to experimentally determine all relevant data in practice. In the DiKey project, highly accurate and broadly applicable models for predicting material data are developed using machine learning (ML) methods, which are much more powerful and flexible than the physical models available so far.
The new forecasting models to be developed will be tailored to the needs of the chemical industry from the beginning. For this purpose, Federated Learning and encryption technologies are being developed and directly implemented in the industry to confidentially incorporate company data into model training. Thus, a very large number of process variants can be examined and evaluated without experimentation.
Simulation of Process Alternatives
Based on the results, the most promising processes will be selected and then optimized using INOSIM Insight Software for Process Simulation. Finally, they will be evaluated regarding their sustainability and economic viability by INOSIM simulation experts. We will combine data-based and knowledge-based methods to identify the best possible interconnection for given synthesis and separation tasks from a portfolio of individual devices. These methods are to be practically tested during the project duration with associated industrial project partners in order to demonstrate the added value of the models. Ultimately, the goal is to determine which new business models for the chemical industry can be built on the insights gained.
DiKey is funded by the Federal Ministry for Economic Affairs and Energy as part of the 8th Energy Research Program. Participants include, under the coordination of Fraunhofer IWTM, DECHEMA e.V., RPTU Kaiserslautern University, Ruhr University Bochum, and INOSIM GmbH. As associated partners, Bayer AG, Evonik Operations GmbH, and Merck KGaA contribute their own material data and application scenarios and test the practical applicability of the project results.
The DiKey project started on April 1, 2025, and will run for three years. During the project, experts will have the opportunity to discuss the findings from the project in stakeholder workshops. Interested? Then register now on the project website.
General Inquiries 
