Optimization of supply air flow and temperature for VAV terminal unit by artificial neural network

Abstract

The purpose of this paper is to optimize the supply air flow rate and temperature of the variable air volume (VAV) terminal unit by artificial neural network (ANN). In general, the setpoint of the VAV terminal unit are supply air flow rate and supply air temperature. These setpoints are determined by the design value, and are calculated by the maximum indoor heating/cooling load and the ventilation requirements. The setpoints calculated based on the design value does not cause any problems in maintaining the indoor environment. However, energy consumption cannot be optimized because it is based on the maximum design value. In order to improve the existing control method, it is important to apply the setpoints according to the real time indoor conditions. The real time indoor condition can be utilized for control through prediction model. In this study, the indoor load, indoor air quality and energy consumption were predicted and used for VAV system control. In addition, the predictive model was developed with the ANN algorithm, and the process of selecting input data and optimizing the predictive model was performed. The performance of the ANN based optimal control algorithm (suggested CASE) for the VAV terminal unit in the target building was compared with those of the dual maximum control algorithm (existing CASE), one of the VAV terminal unit control algorithms. The comparison of heating season showed that the ANN based control algorithm of VAV terminal unit reduced 16.7% of supply fan energy consumption and 19.5% of reheat coil energy consumption compared to the existing CASE using the fixed setpoint.