Convex partition zoner: a new algorithm for automated thermal zoning

This paper introduces a new algorithmic approach to generate perimeter and core thermal zones of energy models from building floor geometries when the actual HVAC zones are unknown or not yet designed, adhering to the ASHRAE standard 90.1 appendix G method. Applicable to early-stage architecture design explorations and urban scale building energy modelling, our prototype is tested on parametrically generated complex floorplans, massing, and a part of Boston's urban footprint, then further compared with a state-of-the-art automation method. Results demonstrate that our prototype can create valid thermal zones from complex and realistic geometry (in 97% of parametric floorplan test cases), is faster (by up to 70% compared with state-of-the-art) and less prone to error, and its simulation results are closer to those of manually zoned energy models (within 0–2% of the manual baseline versus −14% to +26% for state-of-the-art).