Using BEopt (EnergyPlus) with energy audits and surveys to predict actual residential energy usage: Fid-Bau Portal

Using BEopt (EnergyPlus) with energy audits and surveys to predict actual residential energy usage

Rhodes, Joshua D. / Gorman, William H. / Upshaw, Charles R. / Webber, Michael E.

Graphical abstract Display Omitted

Highlights • We build energy models from residential energy audits. • We compare modeled energy use to measured energy use. • The model can be off large amounts for single homes. • The model does well with groups of homes. • Simplified home geometries do not significantly alter model results.

Abstract Simulations of building energy use can give insights into how energy efficiency retrofits and operational changes can influence a building's total and temporal energy use. However, before those models are used to generate recommendations, it is important to understand how accurately the simulations predict actual energy use. This paper seeks to determine model accuracy by considering 54 homes in Austin, TX that are part of a smart grid demonstration project. Data about the homes were collected with energy audits then converted into energy models using BEopt (version 2.0), a residential building-focused graphical user interface (GUI) front-end for EnergyPlus developed by NREL. Actual meter reads (kWh) were compared to simulation results for four cases: (1) energy model output with typical meteorological year (TMY) weather data, (2) output with an actual meteorological year weather (AMY) file created using real weather data from a year during which actual energy use measurements were made, (3) model output with thermostat user behavior data, and (4) output using a simplified home geometry. The purpose of this analysis was to evaluate how well this model, as it is, could be expected to predict energy usage using varying levels of inputs that might be included on an energy audit and survey. Results indicate that the modeling software was able to estimate aggregate annual electrical energy usage within 1% (for groups of homes) but might vary up to 28% (absolute) for an individual home. Using AMY weather data (Case 2) yielded results within 2.4% of measured (aggregate) energy use, while using a TMY3 weather file for Austin yielded results within 9% of (aggregate) actual usage. Using a simplified home geometry did not appreciably change the results of the model or its ability to simulate real usage. Results show that while the model does a reasonable job predicting usage for homes with average usage, the model has trouble predicting usage of homes with low or high consumption, particularly homes below 5kWh/ft²/year (0.47kWh/m²/year) of average energy use or above 10kWh/ft²/year (0.93kWh/m²/year), likely due to unknowns about the home when BEopt defaults were used.