HVAC energy consumption evaluation based prediction of indoor thermal comfort in buildings under artificial neural network (ANN)

• Main Author: Keow, Chin Lun
• Other Authors: Soh Yeng Chai
• Format: Final Year Project
• Language: English
• Published: 2016
• Subjects: DRNTU::Engineering
• Online Access: http://hdl.handle.net/10356/67394
• Institution: Nanyang Technological University

Among the share of energy consumption in building sector, Heating, Ventilating and Air Conditioning (HVAC) systems approximately contribute up to 50% of energy usage. There is a huge energy savings potential if starting with reducing energy consumption of HVAC system by effectively control. The main purpose of this project is to determine the optimal operating points of a Chilled Water Air Conditioning System, which consists of supply air fan in an Air Handling Unit (AHU), compressor pump, condenser fan and water pump in a chiller on condition that thermal comfort for occupants are satisfied. A Levenberg-Marquardt training algorithm based Artificial Neural Network (ANN) approach was adopted to predict Fanger’s Predicted Mean Vote (PMV) value that indicates thermal comfort level. The results show that the predicted PMV values are highly accurate with maximum Mean Absolute Error (MAE) less than 4%. In this report, the dependency of ambient temperature and air velocity on operating frequencies and the dependency of energy consumption on operating frequencies will be discussed. The relationship between them will be formulated by mathematical models, respectively. Then, the energy consumptions for different operating points were evaluated, thus the optimal operating points were found. Finally, the comparison of energy consumptions between optimal operating points and normal operating points at each tolerable thermal conditions are presented and discussed. The results suggest that the energy savings can be up to 42%. Hence, this method can be also applied to other HVAC systems to achieve energy conservation.