Artificial neural network and neuro-fuzzy methodology for phase distribution modeling of a liquid-solid circulating fluidized bed riser

Artificial Neural Network (ANN) and Adaptive Neuro-Fuzzy Inference System (ANFIS) modeling techniques are applied to study the radial and axial solids holdup distributions in a liquid-solid circulating fluidized bed (LSCFB) system. The modeling process is based on the experiments that were conducted using 500 μm size glass beads as solid phase. The radial nonuniformity of the solids holdup is observed under different superficial liquid velocities at superficial solids velocity of 0.95 cm/s and auxiliary liquid velocity of 1.4 cm/s at four axial locations (H = 1.0, 2.0, 3.0, and 3.8 m above the distributor). The effects of different operating parameters such as auxiliary and primary liquid velocities and superficial solids velocity on radial phase distribution in different axial positions of the riser are considered in the model development and analysis. The adequacy of the developed models is investigated by comparing the model predicted and experimental solids holdup data obtained from the pilot scale LSCFB reactor. Radial nonuniformity of the solids holdup is observed under different superficial liquid velocities at superficial solids velocity of 0.95 cm/s and auxiliary liquid velocities of 1.4 cm/s at four axial locations (H = 1, 2, 3, and 3.8 m above the distributor). The cross-sectional average solids holdup in axial directions is compared to the output of the two models. The model outputs show good agreements with the experimental data and reasonable trends of phase distributions. The correlation coefficient values of the predicted output and the experimental data are 0.95 and 0.96 for ANFIS and ANN models, respectively.