Predicting the Residence Time Distribution in Large-Scale Continuous Apparatuses Using Recurrence CFD

In the design of continuous processes, the residence time distribution is a critical criterion. It quantifies both mixing and exposure of the material to the process at hand, thus correlating with product quality and homogeneity. The state of the art is the experimental measurement which is both costly and not suitable for rapid design iterations. Steady-state CFD simulations allow for the numerically cheap calculation of residence time distributions, but neglect inherently transient, but recurrent, mixing patterns. If the patterns occur on a sub-grid length scale, they can either be modelled, as frequently done in RANS simulations or be resolved in transient simulations that are computationally demanding. When the time scale of these mixing patterns is much faster than the time scale of residence times, their resolution by transient simulation may prove too numerically expensive to yield a true residence time distribution. If the length scale of the relevant, recurring mixing pattern is resolved by the grid and by orders of magnitude faster than the mean residence time, neither a transient simulation nor modelling in analogy to RANS are feasible.

Recurrence CFD (rCFD) offers an opportunity to time-extrapolate global recurrent mixing patterns in a physically consistent manner while achieving speedups of up to three orders of magnitude. This enables the calculation of true residence time distributions after capturing the system dynamics from a fully-resolved simulation.

In this study a pilot-scale spouted bed too large to simulate residence time distributions directly is considered, showcasing the utility of the method. CFD-DEM simulations were sampled and time-extrapolated using rCFD to derive residence time distributions. These were compared to experimentally obtained distributions. The influence of additional stabilizing draft plates was assessed to gauge their value in shaping the residence time distribution.