Development and validation of adaptive optimal operation methodology for building HVAC systems

An innovative approach to building operation, called the adaptive optimal operation methodology (AOOM), is developed and validated in this study. The AOOM, which resides in the building energy management and control system, estimates the building and heating, ventilating, and air conditioning (HVAC) system loads and parameters and supplies the local controllers the optimal set points that minimize the energy cost while maintaining occupant comfort. The AOOM uses the recursive least square method with an adaptive forgetting factor to estimate the parameters for the building zones and HVAC systems. A system model is based on the estimated parameters and is used to predict the system dynamic behavior and energy consumption. A genetic algorithm optimizer together with the system model is then used to generate the optimal set points, such as the supply air temperature set point as well as the set points of minimum air flow rate and zone temperature for each zone. The system model is validated through different types of validation experiments. Using the system model with the estimated parameters, the prediction errors are generally less than 0.5°F for the temperatures, are less than 50 cfm for the VAV entering air flow rate, and are less than 5% of the energy value for the energy quantities. Several normalization experiments are examined to improve and quantify the matchness between two full scale HVAC systems. System level validation experiments conducted during the heating and cooling seasons indicate that the HVAC system operated under the AOOM consumes 3 to 10.8% less heated water energy during the heating season and 1 to 4% less electrical energy during the cooling season when compared with a commercial operation methodology. More actuator travel distance is observed by using the AOOM. The AOOM maintains similar occupant comfort conditions as those for the commercial operation methodology.