Using models to provide predicted ranges for building-human interfaces

Most building energy consumption dashboards provide a snapshot of building performance; others provide detailed historical data for comparison to current usage. This paper discusses the Building Agent platform, which was developed and deployed in a campus setting at the National Renewable Energy Laboratory as part of an effort to maintain the aggressive energy performance achieved in newly constructed office buildings and laboratories.  The Building Agent provides aggregated and coherent access to building data, including electric energy, thermal energy, temperatures, humidity, and lighting levels, and occupant comfort feedback, which are displayed in various manners for visitors, occupants, facility managers, and researchers. This paper focuses on the development of visualisations for facility managers, or an energy performance assurance provider, where metered data are used to generate models that provide live predicted ranges of building performance by end use. These ranges provide simple, visual contexts for displayed performance data without requiring users to assess historical information or trends. Several energy modelling techniques were explo red, including static lookup-based performance targets, reduced-order models derived from historical data using main effect variables such as solar radiance for lighting performance, and integrated energy models using a whole-building energy simulation program. This paper describes the new building construction backdrop that has motivated this work, the system architecture providing access to building data, the various modelling approaches currently employed, and the visualisation methods used to display performance and modelled data.