HVAC systems account for 38% of building energy usage. Studies have indicated at least 5-15% waste due to unoccu- pied spaces being conditioned. Our goal is to minimize this waste by retrofitting HVAC systems to enable room-level zoning where each room is conditioned individually based on its occupancy. This will allow only occupied rooms to be conditioned while saving the energy used to condition un- occupied rooms. In order to achieve this goal, the effect of opening or closing air vent registers on room temperatures has to be predicted. Making such a prediction is complicated by the fact that weather has a larger effect on room temper- atures than the settings of air vent registers, making it hard to isolate the influence of the HVAC system. We present a technique for dynamically estimating the heat load due to weather on room temperatures and subtracting it out in or- der to predict the effect of the HVAC system more directly.

Heating, ventilation, and air-conditioning (HVAC) systems use a large amount of energy, and so they are an interesting area for efficiency improvements. The focus here is on the use of semiparametric regression to identify models, which are amenable to analysis and control system design, of HVAC systems. This paper briefly describes two testbeds that we have built on the Berkeley campus for modeling and efficient control of HVAC systems, and we use these testbeds as case studies for system identification. The main contribution of this work is that the use of semiparametric regression allows for the estimation of the heating load from occupancy, equipment, and solar heating using only temperature measurements. These estimates are important for building accurate models as well as designing efficient control schemes, and in our other work we have been able to achieve a reduction in energy consumption on a single room testbed using heating load estimation in conjunction with the learning-based model predictive control (LBMPC) technique. Furthermore, this framework is not restrictive to modeling nonlinear HVAC behavior, because we have been able to use this methodology to create hybrid system models that incorporate such nonlinearities.

HVAC systems account for 38% of building energy usage. Studies have indicated at least 5-15% waste due to unoccu- pied spaces being conditioned. Our goal is to minimize this waste by retrofitting HVAC systems to enable room-level zoning where each room is conditioned individually based on its occupancy. This will allow only occupied rooms to be conditioned while saving the energy used to condition un- occupied rooms. In order to achieve this goal, the effect of opening or closing air vent registers on room temperatures has to be predicted. Making such a prediction is complicated by the fact that weather has a larger effect on room temper- atures than the settings of air vent registers, making it hard to isolate the influence of the HVAC system. We present a technique for dynamically estimating the heat load due to weather on room temperatures and subtracting it out in or- der to predict the effect of the HVAC system more directly.

A significant amount of the world's energy is used for heating, ventilation, and air conditioning (HVAC) systems. This makes them an important target for energy efficiency improvements. One step toward this goal is to find a mathematical model that accurately predicts the performance of current systems. This is difficult, as there is a large variety of HVAC configurations used in both residential and commercial settings. A common energy-reducing setup in residential buildings is the electrical, dual stage heat pump air conditioner. To study this setup, we collect data from a residential testbed which has been outfitted with sensor networks for the purposes of in situ experimentation. Sensors have been deployed which detect the location and activities of occupants. We aim to use this information to develop new, energy-efficient control strategies for the HVAC system at hand. However, these strategies necessitate the use of models which will predict system performance. We develop a two-stage model which, first, learns thermal patterns within the building when the system is OFF due to variable factors such as sunlight, cloud coverage, and wind. This model is then included in the model when the system is ON, allowing us to predict the effect of the system configuration more accurately. Results from this model allow the prediction of temperature within an interval suitable to enable control. This model is scalable to similar systems, and thus can be used to improve the efficiency of HVAC systems by helping to determine more effective control schemes.

A significant amount of the world’s energy is used for heating, ventilation, and air conditioning (HVAC) systems. This makes them an important target for energy efficiency improvements. One step towards these goals is to find a mathematical model that accurately predicts the performance of current systems. This is difficult as there is a large variety of HVAC configurations used in both residential and commercial buildings. One of the most common types of configurations is the Variable Air Volume (VAV) system, which is commonly found in large commercial buildings. To study this setup, we have collected data from the fourth floor of Sutardja Dai Hall, a commercialsized building on the UC Berkeley Campus which has been outfitted with sensor networks.We use semiparametric regression to identify a mathematical model of temperature dynamics based on experimental data. Using this model, we study the effects of occupancy on temperature and isolate certain control variables within the system. The results from this model can be used to improve the efficiency of similar HVAC systems by helping to determine more effective control schemes.