Gazprom Neft has developed an in-house neural network system named Alchemist to optimize the formulation of motor, transmission, and industrial oils. The company opted for internal development because no commercial software existed for this specific task.
The system leverages 15 years of proprietary data collected by the company. By automating the selection process for thousands of base oil and additive combinations, the platform has reduced the time required to bring a new recipe to market from six months to one or two months.
The Alchemist system has already been utilized to refine more than 200 oil formulations. While the company keeps the model architecture and data labeling methods confidential, the project demonstrates how internal data assets can solve complex industrial challenges where off-the-shelf solutions are unavailable.
