Hybrid Dynamic-Empirical Building Energy Modeling Approach for an Existing Campus Building

Abstract

A hybrid modeling framework was constructed to investigate the uncertainty in modeling the energy consumption of an existing campus building with minimal instrumentation. The hybrid framework consisted of a dynamic model of the building’s conditioned spaces, coupled with an empirical model of the building’s HVAC system. The empirical model was calibrated using linear regression of available HVAC system temperature and flow measurements from a building automation system to develop estimates of internal loads and relationships between envelope heat gains/losses and indoor/outdoor temperatures. Crabtree Hall, a 40-year-old building at the University of Pittsburgh, was used as an illustrative case study for this approach. A separate data collection time frame was used for empirical model verification in addition to the initial model development time frame. Comparative results from the model showed a 20% normalized RMS deviation for hourly net heating and cooling for the average day in a given month. This close agreement highlights the possibilities of this approach for rapid assessment of energy consumption and retrofit potential in existing buildings. Future work will include additional refinement of the components of energy consumption using mobile equipment to collect targeted measurements at additional locations, as well as cross-checking existing measurement locations. Additional future work should include extending this method to buildings with more complex HVAC systems, such as variable-air-volume systems and multiple thermal zones, to further verify and improve its robustness.