AI-based multiobjective optimization of flat plate solar collectors for sustainable dairy industry in Mexico

This study develops a computational methodology to assess the performance of a flat plate solar collector system for dairy pasteurization, integrating computational intelligence methods. A heat exchanger tank captures thermal energy from the collectors, with an auxiliary heater compensating for any unmet thermal demand. The analysis considers four climate types—temperate, arid, dry, and tropical—along with four auxiliary fuel options: diesel, fuel oil, natural gas, and LP gas. Design variables include solar field area, heat exchanger tank volume, fuel type, and climate conditions. A dataset is generated by varying these parameters, incorporating governing equations for solar technology, industrial thermal requirements, and regional climate data. This dataset trains a surrogate artificial intelligence (AI) model based on artificial neural networks, where input neurons represent the design variables, and output neurons correspond to economic-environmental indicators: net present value, total life cycle cost, and carbon dioxide emission reduction. To identify optimal system configurations, multi-objective optimization is performed using three different algorithms: particle swarm optimization, genetic algorithms, and the whale optimization algorithm. These algorithms generate Pareto diagrams that facilitate the analysis of trade-offs among the performance indicators. Results indicate economic and environmental feasibility across all climate regions when using diesel as auxiliary fuel, with Jalisco's temperate climate being the most suitable for implementation. This methodology offers a flexible framework for evaluating the feasibility of solar technologies in industrial processes, supporting sustainable energy integration.