Early detection of agglomeration in conical spouted beds using recurrence plots

The agglomeration of particles in a conical spouted bed was investigated using a recurrence plot (RP) and recurrence quantification analysis (RQA) of the pressure fluctuations (PFs) and acoustic emission (AE) signals. Experiments were carried out in a 45° conical spouted bed with sugar particles (d_p = 720 μm; ?_p = 1580 kg/m³). Water was sprayed incrementally into the bed to produce agglomerates during the operation. Several recurrence quantification parameters were calculated during the agglomeration process, and the most suitable ones were chosen for early prediction of the agglomeration in the bed. The results show that recurrence rate, determinism, and laminarity of PFs and AE signals increase during the agglomeration process, which indicate that bed behavior becomes more periodic and deterministic in nature. Additional examination of the RQA parameters show that AE signals are substantially more sensitive to the hydrodynamic changes that occur in the bed, compared to those of PFs, and therefore can detect changes earlier, with more accuracy.