Moisture-induced variations in effective thermal conductivity, diffusivity, and physical properties of fertilizers: an experimental exploration and artificial neural network analysis

Abstract The manufacturing process of phosphate fertilizer demands a substantial amount of thermal energy. Moisture contained in fertilizer particles is a key parameter affecting the performance of the energy transfer systems. For this reason, heat transfer parameters must be precisely known for the process’s characterization, design, and optimization. The conductivity and effective thermal diffusivity of phosphate fertilizers under different moisture contents were determined in this work considering two different approaches: experimental determination by the linear heat probe and Dickerson methods and prediction by a perceptron feed-forward neural network. The results showed that both diffusivity and effective thermal conductivity varied linearly with moisture content within the range of 7.19–8.67 × 10 −8 ± 0.193 m 2 s −1 and 0.172–0.284 ± 0.012 W m −1 K −1 , respectively. The developed neural network demonstrated its ability to accurately predict thermal conductivity, even for conditions outside the training dataset. For a wide range of particle moisture content, the relative deviation was 5.76%. The characterization of the particulate material, based on the size and shape factors of the particles, scanning electron microscopy, specific heat, density, and thermogravimetry, were discussed and reported. The thorough results provided a comprehensive understanding of fertilizers and their behavior in diverse conditions, contributing to the knowledge in this field.