Currently the most commonly used approach to find new chemical formulations and processing parameters is the use of classical Design of Experiments (DoE) software often in combination with statisticians, data scientists or DoE experts. DoE software is often complicated and requires expert knowledge in statistics. This means researchers have to book slots with the limited number of available DoE experts which leads to bottlenecks in the R&D process of many industrial companies.
xT smart_DoE automates this process by combining classical DoE methods with novel AI algorithms. This allows domain experts and laboratory personnel to run experiments without the need of DoE experts in a fast and efficient manner. Compared to other techniques xT smart_DoE finds satisfactory results with less samples required, which additionally reduces R&D time and cost.
Pilot projects with our partner Evonik Industries show strong results in the field of mixture design. In synthetic tests against other state of the art AI algorithms we could show superior behavior, especially for high dimensional problems.