Energy and exergy efficiencies of batch paddy rice drying with liquefied petroleum gas dehumidification: A comprehensive analysis using adaptive neuro-fuzzy inference system and artificial neural networks approaches

• A novel dryer equipped with an LPG-based refrigeration system constructed. • Dehumidification by the LPG system reduced exergy loss and improved drying efficiency. • ANFIS and ANN models were employed to predict drying performance. • The LPG refrigeration system led to an increase in exergy efficiency (0.863) The use of a batch dryer in conjunction with a liquefied petroleum gas (LPG) cooling system has not been extensively implemented due to insufficient research on its effects on rice quality and energy usage. Additionally, there is a scarcity of research focused on analyzing the energy and exergy performance of a batch dryer combined with an LPG refrigeration system for paddy rice. The primary objective of this research is to develop and implement models using artificial neural networks (ANN) and an adaptive neuro-fuzzy inference system (ANFIS) to predict the energy and exergy aspects of paddy rice drying. Energy utilization is calculated on the first law of thermodynamics, while exergy efficiency and loss are assessed using the second law. The factors investigated included drying air temperature (40 °C, 45 °C, 50 °C), air velocity (1.7, 2.2, 2.7 m/s), and relative humidity (70, 80, 90 %) with ambient air dehumidification. Effective moisture diffusivity ranged from 1.51 × 10 −8 to 9.39 × 10 -8 m 2 / s . Specific energy consumption values varied between 21.9786 and 60.7280 MJ / k g , while exergy efficiency ranged from 0.6116 to 0.863. The results demonstrated that the dehumidification of the air played a crucial role in minimizing exergy loss and enhancing overall drying performance. Furthermore, the findings indicated that the ANFIS model exhibited a high coefficient of determination (R 2 ) and a low root mean square error (RMSE), highlighting its strong capability in predicting the drying process. Therefore, based on the data collected, the ANFIS method proved to be more effective in assessing all output parameters compared to the ANN technique.