Calibrating Building Energy Demand Models to Refine Long-Term Energy Planning
A new, flexible calibration approach improved model accuracy in capturing year-to-year changes in building energy demand
Aggregated building energy demand models, which are based on combining the outcomes of many individual building simulations, are an emerging tool for long-term energy planning at multiple spatial scales. They can be used to understand and project changes in building energy demand due to changes in population, climate, and building technologies. However, these models can be hard to calibrate because real-world data availability at the appropriate temporal, spatial, and sectoral scales is often limited. A new approach developed at the U.S. Department of Energy’s (DOE’s) Pacific Northwest National Laboratory (PNNL) allows these aggregate models to be calibrated at multiple scales. Researchers used this new method to calibrate PNNL’s Building ENergy Demand (BEND) aggregate model. Once calibrated, BEND successfully captured year-to-year changes in building energy demand due to changes in weather.
Unlike more traditional statistical methods, physically based aggregate models such as BEND can fully capture the dynamic relationships between hourly building energy demand and population, climate, and building technologies. As a result, these models are valuable tools for understanding multisectoral dynamics. The new approach allows BEND and other aggregate models to be calibrated at the scale at which they will be applied, overcoming a key limitation of this class of models. Models such as BEND will improve model projections of future building energy demand at different scales and refine long-term energy planning through integration with grid operations and resource planning models.
PNNL’s BEND model is one of an emerging class of models designed to capture total and hourly building energy demand resulting from the aggregation of tens to hundreds of thousands of individual building simulations. Historically, these aggregate models have proven difficult to calibrate because there is a limited amount of target data available at relevant space, time, and sectoral scales. Researchers developed and demonstrated a novel approach to calibrate BEND, using approximately 100,000 individual simulations of DOE’s EnergyPlus model, against the best available data at the geographic scale of balancing authorities (electricity management subregions). Once calibrated, BEND captured year-to-year changes in total and peak building energy demand due to variations in weather within these areas. The study applied PNNL’s new calibration approach to the western United States, but the method can be applied to regions across the world with similar data and scale challenges. Researchers also suggested areas in which improved data collection and sharing would help to further refine these emerging models.
Jennie Rice, Pacific Northwest National Laboratory, firstname.lastname@example.org
Ian Kraucunas, Pacific Northwest National Laboratory, email@example.com
This research was supported by the DOE Office of Science as part of research in the MultiSector Dynamics, Earth and Environmental System Modeling Program. A portion of the research was performed using PNNL’s Institutional Computing resources.
Revised: October 22, 2019 | Published: June 6, 2019
Taylor ZT, Y Xie, CD Burleyson, N Voisin, and I Kraucunas. 2019. “A multi-scale calibration approach for process-oriented aggregated building energy demand models.” Energy and Buildings 191:82‒94. https://doi.org/10.1016/j.enbuild.2019.02.018.