Hybrid phenomenological/ANN-PSO modelling of a deformable material in spouted bed drying process

Abstract

In this work, a hybrid (phenomenological/ANN-PSO) model has been developed to simulate the spouted bed drying of deformable solid materials, considering material shrinkage and the physical property variation during drying. Accordingly, an artificial neural network (ANN) model has been coupled to a phenomenological one to describe the heat and mass transfer during the drying of these materials, specifically of guava pieces, in a spouted bed dryer. The optimum architecture of ANN (4–7-3) has been obtained using a Particle Swarm Optimisation (PSO) algorithm. This model demonstrated higher accuracy in its ability to estimate the material physical properties (R2 = 0.99, MSE = 0.00048 and RMSE = 0.069). Furthermore, a comparison between the model results and experimental data provided high correlation. This differs from the usual approach, which neglects variation of the physical properties; the hybrid model is able to simulate the drying deformable particle process behaviour considering the transient variation of the properties obtained from the ANN-PSO model. The results differ significantly from those predicted with the assumption of constant properties.