Scientific Reports Article

Intelligent multi-objective optimization of thermal comfort and ventilation performance in stratum ventilation design

Authors: Nadia Ghezaiel Hammouda, Zakarya Ahmed, Ihab Omar, As'ad Alizadeh, Narinderjit Singh Sawaran Singh, Borhen Louhichi, Walid Aich & Banafshe Hamidi

Abstract: Stratum ventilation (SV) has emerged as a promising approach for simultaneously addressing indoor thermal comfort, airflow effectiveness, and energy efficiency. Yet, most prior research considers predictive modeling, optimization, and decision-support separately, which reduces their usefulness in practice. To overcome this gap, the present study develops an integrated hybrid framework that links machine learning models, metaheuristic optimization, and multi-criteria decision-making into a unified workflow for SV enhancement. The proposed methodology unfolds in four sequential phases: (1) data preparation and statistical assessment, (2) development of predictive models using artificial neural networks (ANN) optimized through genetic algorithm (GA) and leader Harris hawks optimization (LHHO), (3) multi-objective optimization employing NSGA-III, and (4) ranking of Pareto-optimal solutions with the VIKOR method to accommodate different operational priorities. The findings indicate that GA-assisted ANN consistently achieved superior prediction accuracy (R > 0.995) compared to LHHO-ANN. Optimal thermal comfort was obtained with supply air velocities of 1.18–1.20 m/s, supply air temperatures around 22.0–22.2 °C, and clothing insulation levels near 1.0 clo. Ventilation performance benefited from small vane angles (≤5°) and cooler wall surface temperatures (≤12 °C), while stratification was mitigated under wider vane angles (>10°) combined with moderately higher wall surface temperatures (13–14 °C). Heating efficiency proved robust across all candidate solutions, with a consistent utilization coefficient of approximately 1.58. The VIKOR-based ranking organized the Pareto front into ten representative design scenarios, each offering a balanced trade-off among comfort, air quality, and energy use under varying preference weights. By structuring prediction, optimization, and decision-making in a single framework, this study delivers actionable strategies for tailoring SV operation in diverse settings such as office buildings emphasizing comfort, healthcare spaces requiring ventilation effectiveness, and large halls where stratification control is critical.