Uncertainty and sensitivity analysis of spatio-temporal occupant behaviors on residential building energy usage utilizing Karhunen-Loève expansion

• Published in: Building and environment 2017-04, Vol.115, p.157-172
• Main Authors: O'Neill, Zheng, Niu, Fuxin
• Format: Article
• Language: English
• Subjects: Behavior; Building construction; Buildings; Case studies; Coefficient of variation; Computer simulation; Detaching; Electricity; Electricity consumption; Energy consumption; Energy policy; Energy usage; Federal agencies; HVAC; Illumination; Karhunen-Loève expansion; Lighting; Mathematical models; Monte Carlo simulation; Occupant behaviors; Perturbation methods; Residential buildings; Residential energy; Schedules; Sensitivity analysis; Spatio-temporal; Thermal expansion; Thermostats; Uncertainty; Uncertainty analysis; Uncertainty and sensitivity analysis
• ISSN: 0360-1323; 1873-684X
• DOI: 10.1016/j.buildenv.2017.01.025

Occupant behaviors (e.g., switch/dim lights, adjust thermostat, etc.) have a big impact on building energy consumption. However, the occupant behavior in buildings is not well understood and oversimplified. In addition, a majority of current researches just focused on the occupant number variation and ignored the temporal variation (e.g., the occupant presence duration and arrive/leave time). This paper aims to use a whole building simulation program (i.e., EnergyPlus) to investigate the impacts of spatio-temporal occupant behaviors on residential building energy usage utilizing a simple static schedule-based behavior model through uncertainty analysis (UA) and sensitivity analysis (SA). A residential prototype building EnergyPlus model with two thermal zones from the U.S. Department of Energy is used for the case study. The chosen building type is a single-family detached house with three locations in U.S. 3000 EnergyPlus Monte Carlo simulations using a Karhunen-Loève (KL) expansion based sampling method for each location is conducted. Occupancy behaviors related parameters including thermostat setpoints, occupancy and lighting schedules are randomly perturbed using a KL truncated expansion, which results in a 3.76% overall average coefficient of variation of selected input parameters. UA with such perturbation shows about a 4% impact on the HVAC annual and peak energy consumption, Annual building and HVAC electricity consumption, and hourly peak building and HVAC electricity consumption are selected for the study of impacts of the spatio-temporal occupant behaviors on building energy usage using Sobol and PEAR sensitivity index. SA Results show that thermostats setpoints have big influences on residential building energy usage. •EnergyPlus Monte Carlo simulations.•A Karhunen-Loève truncated expansion based sampling method.•Spatio-temporal occupant behaviors investigated using Sobol sensitivity index.•A meta-model based on a Bayesian network model.•Thermostats setpoints have big influences on residential building energy usage.