The more the better? Archetype segmentation in urban building energy modelling

Abstract

Urban building energy modelling is gaining traction as a planning tool to support widespread decarbonization of the built environment. Building-scale models allow for the evaluation of specific emission reduction policies at an urban scale. Given the limited availability of building-by-building data on construction standard and program, aggregating building information through archetypes is key, but a poorly understood step in the urban energy modelling process. In this study, different levels of archetype segmentation are explored for the city of Oshkosh, WI (∼13,000 buildings). A comparison of actual, city-level energy with UBEM simulations suggests higher levels of archetype segmentation do not necessarily lead to higher accuracy, leading to models that are both accurate and nimble enough to explore a variety of upgrade scenarios. Informing archetypal segmentation with policy-informed metrics is beneficial, but pursuing increased detail could dangerously reduce accuracy without ground-truth data.