Prediction of plug loads in office buildings: Simplified and probabilistic methods

To predict buildings' energy use, multiple systems and processes must be considered. Next to factors such as building fabric and construction, indoor environmental control systems, and weather conditions, the energy demand attributable to buildings' internal heat gains resulting from inhabitants, lighting, and equipment usage also needs to be addressed. Given this background, the present contribution focuses on plug loads in office buildings associated mainly with computers and peripherals. Using long-term observational data obtained from a continuously monitored office building in Vienna, we specifically explore the relationship between inhabitants' presence, installed power for equipment, and the resulting electrical energy use. The findings facilitate the formulation of both simplified and probabilistic office plug loads predictions methods. Thereby, the model evaluation results suggest that the non-stochastic model provides fairly reasonable predictions of annual energy use associated with plug loads. However, the stochastic plug load model - together with a stochastic occupancy model - outperforms the simplified model in predicting the plug loads peak and distribution.