Ontology-driven digital twins for optimized recommissioning of hospital building operations

Optimizing setpoints in non-residential buildings is complex due to multiple competing objectives, such as: energy efficiency, occupant comfort, and cost. This paper presents a multi-objective optimization framework integrated with an ontology-based digital twin for building operation optimization. Using the Non-Dominated Sorting Genetic Algorithm II, the framework balances thermal discomfort, CO₂ levels, and energy costs, using semantic ontology models for building topology definition and component constraints. Applied to a hospital case study, the approach reduced thermal discomfort by 92%, quantified using kelvin-hours, and reduced operational energy costs associated with space heating and ventilation-related electricity consumption by 48%, relative to baseline operation. These results were obtained using a comfort-focused optimal strategy with 73 decision variables selected from the Pareto front. Robustness was confirmed for previously unseen operational periods that exhibited variations in dynamic factors, such as occupancy and weather conditions. Sensitivity analysis identified space temperature setpoints as primary optimization drivers, followed by supplied air temperature. This scalable framework supports building management system recommissioning by mapping results to controllers, suitable for diverse non-residential buildings.