The multiphase simulations of gas-solid fluidized beds are limited to short durations because they are computationally expensive and require huge simulation support. This has made the gas-solid process too slow to picture particle conversion or heating of solid. However, numerical models such as CFD-DEM and TFM showed consistent performance towards lab-scale simulations, but applying these methods to plant-scale with complex geometries is still challenging. These industrial processes involve complex modelling varying from length to time scale, limiting the classical CFD(-DEM) simulations to short durations. Although plant-size reactors can be simulated using these methods, they are limited by short-term investigations because of the enormous computational power required.

So, time scales are very challenging to picture in large-scale fluidized beds, which take hours or days to simulate the relevant dynamics involved within the system. Therefore, an efficient time-extrapolation of numerical simulations is regarded as an evident step in solving long-term processes ranging from industrial plant scale to industrial process scale. Data-based recurrence CFD can be a promising candidate for studying pseudo-periodic processes by extrapolating globally recurring patterns using the Eulerian approach and solving the transport and interaction of passive scalars in the Lagrangian framework.

This research project aims to develop further data-based recurrence CFD (rCFD) to study the species conversion process of pseudo-periodic systems to lab-scale and pilot-scale fluidized beds. We mainly studied the outgoing secondary gas concentration and solid mixing profile with varying physical diffusion coefficient terms to compare the rCFD results with classical CFD simulations. We also performed 3D calibration by introducing a checkerboard pattern specifically focusing on studying local diffusion within the domain. We replicated the rCFD simulations by reducing the computational time to more than two orders of magnitude, capable of matching up to the real-time.