The present work proposes a novel radially cross-flow multistage solid–liquid circulating fluidized bed (SLCFB). The SLCFB primarily consists of a single multistage column (having an inner diameter of 100 mm and length of 1.40 m), which is divided into two sections wherein both the steps of utilization or loading (e.g., adsorption and catalytic reaction) and regeneration of the solid phase can be carried out simultaneously in continuous mode. The hydrodynamic characteristics were studied using ion exchange resin as the solid phase and water as the fluidizing medium. The loading and flooding states were determined for three particle sizes; i.e., 0.30, 0.42, and 0.61 mm. The effects of the superficial liquid velocity and solid feed rate on the solid hold-up were investigated under loading and flooding conditions. The solid hold-up increases with an increase in the solid feed rate and decreases with an increase in the superficial liquid velocity. An artificial-intelligence formalism, namely the multilayer perceptron neural network (MLPNN), was employed for the prediction of the solid hold-up. The input space of MLPNN-based model consists of four parameters, representing operating and system parameters of the proposed SLCFB. The developed MLPNN-based model has excellent prediction accuracy and generalization capability.

目前的工作提出了一种新颖的径向错流多级固液循环流化床（SLCFB）。SLCFB主要由一个多级塔（内径为100 mm，长度为1.40 m）组成，该塔分为两部分，其中利用或装载（例如，吸附和催化反应）和再生的步骤固相可以在连续模式下同时进行。以离子交换树脂为固相，水为流态化介质，研究了水动力特性。确定了三种粒径的加载状态和溢流状态。即0.30、0.42和0.61毫米。研究了在加载和驱油条件下表观液体速度和固体进料速率对固体滞留率的影响。固体滞留率随固体进料速率的增加而增加，随表观液体速度的增加而降低。人工智能形式主义，即多层感知器神经网络（MLPNN），被用于预测固体含量。基于MLPNN的模型的输入空间由四个参数组成，分别代表所提出的SLCFB的操作系统和系统参数。所开发的基于MLPNN的模型具有出色的预测准确性和泛化能力。代表建议的SLCFB的操作系统和系统参数。所开发的基于MLPNN的模型具有出色的预测准确性和泛化能力。代表建议的SLCFB的操作系统和系统参数。所开发的基于MLPNN的模型具有出色的预测准确性和泛化能力。