Main content area

Energy saving estimation for plug and lighting load using occupancy analysis

Anand, Prashant, Cheong, David, Sekhar, Chandra, Santamouris, Mattheos, Kondepudi, Sekhar
Renewable energy 2019 v.143 pp. 1143-1161
algorithms, computers, decision making, energy, energy conservation, lighting, models, prediction, renewable energy sources
The gap between the actual and intended energy use for a building is often attributed to stochastic behaviour of occupants. This study systematically investigates the relationship of occupancy with plug and lighting loads energy consumption for several spaces of an institutional building floor. A new parameter ‘Energy-use per person (K)’ is introduced to explain the stochastic relationship between Energy and Occupancy. A model for K is developed as a function of occupancy using ‘Multiple non-linear regression (MNLR)’ and ‘Deep neural network (DNN)’ based algorithms. DNN algorithm shows a better prediction of K with less Mean absolute percentage error (MAPE) of 9.67% and 2.37% compared to 10.34% and 3.15% of MNLR for plug and lighting loads respectively. The model developed is used to estimate possible energy savings during occupied hours with a rule-based energy-use behaviour. Possible plug load energy savings are 8.9%, 3.1% and 1.3% for the classroom, open office, and computer room respectively. Similarly, possible lighting load energy savings are 65.1%, 43.6% and 38.4% for the classroom, open office, and computer room respectively. The study outcome, a robust and iterative ‘K model’ development process can be used as a support tool in decision making for facility management.