Development and assessment of speed-up algorithms for the reactive CFD–DEM simulation of fluidized bed reactors

Abstract

In this work, we propose a particle agglomeration (PA) speed-up algorithm to reduce the computational cost of the CFD–DEM framework based on the coupled solution (CP) of the gas–solid transport and heterogeneous reaction at the level of each particle in a fluidized bed reactor. The performances of PA are assessed and compared with the ones provided by the previously proposed operator-splitting (OS) and in situ adaptive tabulation (ISAT) speed-up algorithm. The analysis is carried out on methanation and steam reforming systems to investigate different reactions and transport characteristic times under different operating conditions. The results revealed that OS & ISAT is generally computationally more efficient by providing a higher chemical speed-up than CP & PA both in methanation (15 vs. 6) and steam reforming (13 vs. 4). However, OS requires a smaller simulation time step with respect to the CP to achieve convergent results when fast transport/reaction phenomena are considered. Hence, the overall computational cost might be even larger with respect to the CP despite the ISAT technique. Consequently, the selection of the best performing algorithm is not trivial and strongly depends on the characteristic times of transport and reaction which are a function of the local conditions experienced in the bed. Thus, we propose a strategy to select the most adequate algorithm (i.e., CP & PA or OS & ISAT). The strategy is based on the simulation of a test reactor characterized by a small computational cost but conceived to be representative of the chemical and fluid dynamic behavior of the target reactor. This enabled the simulation of a million particle methanation reactor experimentally investigated in the literature. A good agreement between simulation and experimental outlet concentrations (error up to 7% for the main species) is observed along with a significant 20-fold chemical speed-up.