Simulation-driven optimization of direct solar dryers for household use: A combined CFD and ANN-GA approach

Abstract

This study introduces a novel, integrated optimization framework for domestic solar dryers that uniquely combines computational fluid dynamics (CFD), artificial neural networks (ANN), and genetic algorithms (GA) to achieve superior thermal uniformity and enhanced drying performance. Unlike conventional trial-and-error or replication-based designs—which often result in non-uniform temperature fields and inefficient energy usage—this research systematically addresses heat distribution challenges through a data-driven and simulation-validated approach. CFD simulations, conducted using OpenFOAM and validated via no-load experimental testing, revealed non-uniform drying patterns during initial trials with pineapple slices. These findings informed the development of a machine learning model, where a validated CFD dataset (error <7.33 %) was used to train an ANN-GA system. This hybrid model achieved high predictive accuracy (R² = 0.98) with an average error of only 3.87 %, enabling precise prediction and optimization of dryer performance. The optimized configuration delivered an exceptionally uniform temperature distribution (mean 46.15 °C, SD = 0.07 °C), making a significant advancement over conventional designs. The integration of CFD-based physical modeling with AI-driven optimization constitutes a key innovation of this study, offering a replicable and scalable method for the development of high-efficiency domestic solar drying systems.