Modelling of industrial packed bed reactors: connecting CFD with imaging and AI

Packed bed reactors are pivotal for the energy transition, yet their industrial design and scale-up remain constrained by the wide separation of length and time scales governing transport, reaction and phase behaviour. While high-fidelity pore-scale models, such as particle-resolved computational fluid dynamics (PRCFD), offer unprecedented insights, their high computational cost makes them impractical for industrial design. Consequently, design of large-scale packed bed reactors still relies on empirical correlations that oversimplify multiphase and reactive transport, often resulting in overdesign associated with elevated costs and increased emissions. A substantial gap exists between PRCFD studies of idealised packed beds and the complex environment of real industrial reactors. This perspective article explores the recent trends in PRCFD and positions it in the rapidly evolving modelling domain for packed bed reactors. We further highlight the recent innovations in Reduced Order Modelling (numerical homogenisation, pore network modelling), artificial intelligence and non-invasive imaging techniques, and their integration with PRCFD.