Towards Sustainable Cold Storage: A Hybrid CFD–ML Approach to Temperature and Energy Optimization

Abstract

This research presents an innovative hybrid computational framework that integrates Computational Fluid Dynamics (CFD) with Machine Learning (ML) algorithms to optimize temperature distribution and energy efficiency in cold storage facilities. The study addresses critical challenges in maintaining uniform thermal conditions while minimizing operational costs through advanced modeling techniques. A comprehensive CFD validation study demonstrates excellent agreement with established benchmarks, achieving correlation coefficients exceeding 0.95 for airflow distribution patterns. The hybrid approach employs systematic parameter analysis including surface velocity effects, temporal temperature variations, and multi-zone optimization strategies. Machine learning models trained on extensive CFD datasets achieve remarkable performance improvements, with temperature prediction accuracies reaching R² = 0.94 and energy consumption forecasting achieving R² = 0.91. Comparative analysis between optimized and conventional cold storage operations reveals significant improvements across multiple performance metrics. The optimized system demonstrates 23% reduction in energy consumption, 35% improvement in temperature uniformity, 28% decrease in product weight loss, and substantial reduction in transpiration rates. The framework successfully identifies optimal operational conditions including airflow velocities between 1.2-1.8 m/s and strategic evaporator positioning that enhances thermal performance. This integrated methodology provides a computationally efficient alternative to traditional approaches while maintaining high accuracy, enabling real-time optimization and intelligent control of cold storage systems.