Design and Optimization of a New Passive Building Cooling System Using Water Evaporation: A Numerical and Experimental Approach Guided by Artificial Intelligence

Abstract

This thesis presents the design, optimization, and experimental validation of a passive evaporative

cooling system for buildings, utilizing natural plant fibers, specifically cotton and sisal as eco- friendly evaporative pads. The study aims to provide a sustainable, low-energy alternative to

conventional cooling technologies by combining traditional passive cooling principles with modern analytical tools and artificial intelligence. Two system configurations were evaluated: a basic prototype and an improved model featuring enhanced thermal insulation, water recovery, and real-time performance monitoring. Experimental tests were conducted under varying airflow rates and pad thicknesses to assess thermal performance, cooling effectiveness, and water consumption. Cotton demonstrated superior water retention and cooling performance, while sisal offered greater airflow permeability and structural durability. Numerical analysis using Python and psychrometric modeling enabled detailed evaluation of heat and mass transfer within the system. Additionally, preliminary AI models based on Random Forest regression were developed to predict system behavior under changing conditions, supporting the feasibility of intelligent, adaptive cooling solutions. The results confirm that material properties, pad geometry, and airflow significantly affect system efficiency. The final prototype represents a low-cost, scalable solution for passive building cooling, with potential for integration into smart building infrastructures. This work contributes to the advancement of environmentally responsible cooling technologies and lays the groundwork for future AI-driven passive systems.