Modeling and optimization of orange peel drying using thin-layer equations and artificial neural networks for standardized powder production

Abstract

Globally, about 32 million tons of nutrient-rich orange peels are wasted annually due to the lack of optimized drying methods for efficient preservation and utilization. The present study addresses this gap by investigating the drying kinetics of orange peels were studied under convective hot air drying (CHAD, 90 °C, 5 h) and microwave drying (MD, 180 W, 75 min) to optimize process parameters and improve the quality of dried powders for high-value applications. The results showed that MD significantly decreased drying time compared to CHAD. The drying data were analyzed using five thin-layer models. The Wang & Singh model best fitting the MD data and the logarithmic model best fitting the CHAD data. Additionally, moisture ratio predicted using a multi-layer feedforward artificial neural network (ANN) with backpropagation yielded high R² values for CHAD and MD, confirming the accuracy of model. Importantly, MD allowed superior retention of color and antioxidant properties of orange peels compared with CHAD, while requiring shorter drying time. This study presents a practical approach to sustainably valorizing citrus waste by developing optimized drying protocols that integrate experimental kinetics with machine learning predictions.