Modelling Fluidization by Recurrence CFD

Description

Over the last decades, the numerical modelling of fluidized bed processes has become feasible even for industrial processes. Commonly, continuous two-fluid models are applied to describe large scale fluidization.Too allow for coarse grids novel two-fluid models account for unresolved sub-grid heterogeneities. However, computational efforts remain high – in the order of several hours of compute-time for a couple of seconds of real-time – thus preventing the representation of long-lasting heating or conversion processes. In order to overcome this limitation, data-based recurrence CFD (rCFD) has been put forward in recent years. rCFD can be regarded as a data-based method, which relies on the numerical predictions of a conventional short-term simulation. This data is stored into a database and then used by rCFD to efficiently time-extrapolate the fluidization process in high spatial resolution. Previous studies report on computational speed-ups in the order of three to four orders of magnitude, eventually allowing for real-time simulations of fluidized beds. In this study, we apply rCFD to a set of fluidized beds of different sizes. In a series of short-term simulations, we will compare the numerical predictions of rCFD with those of corresponding expensive full CFD reference simulations. For this comparison we will focus on solid mixing, secondary gas washout, particle size segregation. We further present a sensitivity study on how much rCFD predictions vary with modelling parameters and discuss on how rCFD could be calibrated in order to better agree with full CFD.

Period	06 Jul 2022
Event title	10th International Conference on Conveying and Handling of Particulate Solids
Event type	Conference
Location	ItalyShow on map