Presentation
Unlocking mixing improvements by AI driven geometry optimisation
Time & Location
Thursday: 13.00 to 13.15, Stage 4
Speaker
Dominik WernerCEO - EvoPhase
About this presentation
In industrial processes, effective mixing—characterised by being fast, reliable, and requiring low energy—is essential. The standard method of improving mixing in industrial process equipment primarily involves adjusting process parameters, such as rotation rate or fill height. However, the geometries of the equipment used today have remained largely unchanged for 70 years. This research uses HARPPP, an AI-based geometry optimisation tool from EvoPhase, to optimise the geometry of a ploughshare mixer in laminar flow conditions for enhanced axial mixing. The process involved running 48 parallel simulations, each evaluating the mixing efficiency of different impeller designs over a set period. The AI tool employed evolutionary algorithms to evolve the impeller shape, seeking the optimal design for maximum mixing with constant material properties and rotation rate. The AI identified a double-helix impeller shape, outperforming the initial design by achieving 1000 times better mixing without increasing power requirements. This finding not only enhances the efficiency of mixing processes but also has broad implications for various industrial applications where mixing is essential.
Speaker Bio:
Dominik Werner is a PhD candidate at the University of Birmingham and the CEO and Co-Founder of EvoPhase. His research in fluidised bed reactors and granular materials, employing advanced experimental techniques and modern simulation approaches, has established him as an expert in these domains. Mr. Werner’s academic contributions include publishing and co-authoring over 10 research papers across various fields, alongside presenting his work at multiple international conferences. Recently, Mr. Werner achieved a significant milestone by securing the ICURe Explore Programme, a £30,000 initiative designed to facilitate the commercialisation of research conducted by his team. This achievement has allowed EvoPhase to emerge. The company is now planning to launch its initial product, a software for the automatic optimisation of geometries using simulations.