A new project will replace the slow trial-and-error hunt for battery materials with a self-driving laboratory that designs, makes, and tests chemicals on its own. The problem is that developing better materials for energy storage—such as the chemicals inside flow batteries—currently takes years of manual experimentation. PREDICTOR aims to compress that timeline by combining computer simulations, automated chemical synthesis, and artificial intelligence into a single, rapid pipeline. The AI will learn from each experiment and adjust the next one, creating a feedback loop that speeds up discovery without human intervention. If the system works, it could dramatically accelerate the development of grid-scale batteries that store renewable energy from wind and solar. Flow batteries are a promising technology for this, but they need cheap, stable, and non-toxic active materials to become commercially viable. PREDICTOR’s high-throughput method could identify such materials in months rather than years. The project will validate its approach by building and testing three prototype flow battery cells. Beyond energy storage, the same automated screening platform could be adapted to find materials for other applications, from catalysts to pharmaceuticals.
View original technical description
PREDICTOR aims to establish a rapid, high-throughput method to identify and develop materials for electrochemical energy storage. This method will comprise: A modelling and simulation tool for the computational screening of organic chemicals based on their potential performance in energy storage systems. Automated chemical synthesis, electrolyte production and characterization methods, so that the chemicals identified in the screening step can be rapidly produced and tested for their suitability in energy storage applications. Artificial-intelligence-based self-optimization methods that allow experimental data from material characterization to be fed back into automated experimental methods to enable self-driving laboratory laboratory platforms and for modelling and simulation tools, improving their accuracy. Data management systems to standardize and store the data generated for further use in model validation and self-optimization procedures. This approach will allow the rapid identification, synthesis and characterization of materials within a coherent development chain, replacing conventional trial-and-error developments. It will exploit the synergies between several emerging markets (digital technologies, artificial intelligence, high-throughput experimentation, renewable energy storage), providing the recruited doctoral candidates (DCs) with a valuable interdisciplinary skillset. To validate the PREDICTOR system, the case study will be active materials and electrolytes for redox-flow batteries. Within the project, three demonstrator battery cells (TRL3-4) will be assembled and tested with the newly developed materials.
Plain English summaries and category classifications on this site are generated by AI and may not perfectly reflect the original research.
Is something wrong? Let us know