• Proposes a target-based retrofit framework for multi-zone school buildings • Optimizes energy use, indoor CO 2 levels, comfort, and retrofit costs • Integrates IDA-ICE simulations, surrogates, and NSGA-III optimization • Constrains worst-performing rooms instead of building-wide averages • Identifies nZEB-compliant solutions with CO 2 < 800 ppm and PPD ≤ 10% Retrofit design for nearly zero emission (nZEB) school buildings must reconcile strict energy targets with healthy, comfortable indoor environments. Most multi-objective retrofit studies either model a single representative zone or optimise building-average indicators, which can mask rooms with poor indoor air quality (IAQ) or thermal comfort. In a Norwegian agricultural school, baseline simulations show that building-average CO₂ and predicted percentage dissatisfied (PPD) meet common design targets, while several rooms exceed IAQ and comfort thresholds.We propose a target-based multi-objective optimisation framework that explicitly constrains the worst-performing rooms. Four objectives are considered: annual specific energy use, maximum room CO₂, maximum room PPD, and total retrofit investment cost. Energy use is constrained to the Norwegian nZEB limit for schools (≤ 91 kWh/(m²·y)), with CO₂ and PPD limited to 800 ppm and 10 %, in line with EN 16798 and recent IAQ guidance. Eight design variables (envelope U-values, supply-air temperature setpoints, infiltration, and ventilation rates) are explored using a Definitive Screening Design with 17 IDA-ICE simulations and second-order surrogate models (R² ≥ 0.99). These feed an NSGA-III algorithm to generate a Pareto front, which is post-processed with an Entropy–TOPSIS method to rank solutions. Results show that retrofit strategies can keep every room below 800 ppm CO₂ and 10 % PPD while satisfying the nZEB energy limit at moderate investment levels. The framework is generic and transferable, providing transparent decision support that shifts from building-average metrics toward room-specific occupant protection.
Aganović et al. (Sun,) studied this question.