Predictive modeling of building energy consumption and thermal comfort for decarbonization in construction and retrofitting

• Innovative approach to modeling building energy and thermal comfort for decarbonization. • Integrates 21 key parameters to forecast energy behaviors and optimize retrofit strategies. • Uses EQiTC metric to ensure energy-saving measures do not compromise occupant comfort. • Identifies critical factors in energy efficiency and thermal comfort for targeted retrofits. This study introduces an integrated predictive modeling framework for assessing building energy consumption and indoor thermal comfort, with a focus on supporting decarbonization efforts in both new construction and retrofit scenarios. A total of 21 critical design and operational parameters were evaluated using Monte Carlo simulations combined with EnergyPlus, enabling high-resolution analysis of cooling loads, thermal comfort performance, and retrofit outcomes. The proposed multi-variable regression model demonstrated strong predictive accuracy, achieving an R² of 0.98, a mean absolute percentage error of 1.59 %, and a Coefficient of Variation of the Root Mean Square Error (CVRMSE) of 1.47 % in forecasting annual cooling demands. Optimization of variables such as indoor temperature set-point, solar heat gain coefficient, and glazing U-values yielded energy savings of up to 70 kWh/m² annually, corresponding to a potential carbon emission reduction of 31.5 kg CO₂/m²/year, based on a regional electricity emission factor of 0.45 kg CO₂/kWh. The environmental quality thermal comfort index developed within this framework effectively quantified comfort conditions across varying scenarios, with values improving from 50 to 100 under optimized configurations. The model also revealed pronounced spatial variability, with perimeter zones reaching peak cooling loads of up to 198 W/m², emphasizing the need for zone-specific design strategies.