building management system, model predictive control, genetic algorithm, energy optimization
The rapid improvement of living standards has led to increased energy consumption in buildings worldwide. Globally, the energy consumed in buildings accounts for 20.1% of total delivered energy (EIA 2016). Improving energy efficiency in buildings therefore is an important component for combating climate change. This paper aims to improve end use energy efficiency in multi-zoned residential buildings through the application of thermal comfort based, energy optimization algorithms. We use a case study approach with a detailed analysis of a 4-story residential apartment building in central Illinois. The study building constitutes 21 thermal zones modeled in EnergyPlus. The model is validated using monthly energy consumption data. The effectiveness of four different steam heating system control methods are evaluated and described: a) a Model Predictive Controller (MPC) design based on neuro-fuzzy temperature predictor; b) a Proportional-Integral-Derivative (PID) tuned by fuzzy logic; c) a PID tuned by a genetic algorithm; and d) an on/off controller and the flow regulator based on indoor temperature. All are optimized for energy consumption reduction potential and thermal comfort. The main effect of the various control methods is tuning boiler feed flow by regulating the condensing cycle. A reduction in circulated steam flow results in decreased direct energy consumption and improved condensing pump efficiencies. We find that the MPC design using a neurofuzzy temperature predictor can reduce heating energy use by up to 38% in comparison with an on/off controller baseline.