TY - JOUR
T1 - Numerical modeling of non-uniform indoor temperature distribution for coordinated air flow control
AU - Li, Yuming
AU - Pan, Yiqun
AU - Huang, Zhizhong
AU - Fu, Ling
AU - Li, Jing
AU - Sun, Tianrui
AU - Zhu, Mingya
AU - Yuan, Xiaolei
N1 - Publisher Copyright:
© 2023
PY - 2024/4/1
Y1 - 2024/4/1
N2 - The indoor thermal environment is usually non-uniformly distributed in parameters such as temperature and velocity. For precise control of indoor environment, it is necessary to figure out the indoor non-uniform temperature distribution (INUTD) that empowers independently zone control according to the different requirements of occupants. Computational Fluid Dynamics (CFD) tool is usually used to obtain the INUTD, but it consumes huge computational resources and time, which may not be suitable for coupling with control system. To address this research question, this study proposes a numerical modeling method for predicting the INUTD and air flowrate response of multiple diffusers for coordinated air flow control. Firstly, we establish a dataset with 5000 cases, along with the results of INUTD for each case. Then two models, namely room thermal response model (RTRModel) and air flowrate prediction model (AFPModel), are developed by machine learning algorithms to predict indoor temperature and supply air flowrate, respectively. The results show that proposed models are both fast in prediction, less than 1 s for each case. The standard deviation of error in RTRModel developed with support vector machine algorithm is 0.0041 while that in AFPModel developed with Convolutional Neural Network algorithm is 0.0198. Further, a comparative analysis has been conducted between the AFPModel with and without optimization algorithm. The results reveal that using optimization algorithm is more accurate, but it takes more time. While numerical modeling can instantaneous response with qualified accuracy. The proposed method can contribute to independently zonal environment control and occupant-centered micro-environment control.
AB - The indoor thermal environment is usually non-uniformly distributed in parameters such as temperature and velocity. For precise control of indoor environment, it is necessary to figure out the indoor non-uniform temperature distribution (INUTD) that empowers independently zone control according to the different requirements of occupants. Computational Fluid Dynamics (CFD) tool is usually used to obtain the INUTD, but it consumes huge computational resources and time, which may not be suitable for coupling with control system. To address this research question, this study proposes a numerical modeling method for predicting the INUTD and air flowrate response of multiple diffusers for coordinated air flow control. Firstly, we establish a dataset with 5000 cases, along with the results of INUTD for each case. Then two models, namely room thermal response model (RTRModel) and air flowrate prediction model (AFPModel), are developed by machine learning algorithms to predict indoor temperature and supply air flowrate, respectively. The results show that proposed models are both fast in prediction, less than 1 s for each case. The standard deviation of error in RTRModel developed with support vector machine algorithm is 0.0041 while that in AFPModel developed with Convolutional Neural Network algorithm is 0.0198. Further, a comparative analysis has been conducted between the AFPModel with and without optimization algorithm. The results reveal that using optimization algorithm is more accurate, but it takes more time. While numerical modeling can instantaneous response with qualified accuracy. The proposed method can contribute to independently zonal environment control and occupant-centered micro-environment control.
KW - Computational fluid dynamics
KW - Coordinated air flow control
KW - Machine learning
KW - Non-uniform indoor temperature distribution
UR - http://www.scopus.com/inward/record.url?scp=85179886001&partnerID=8YFLogxK
U2 - 10.1016/j.jobe.2023.108246
DO - 10.1016/j.jobe.2023.108246
M3 - Article
AN - SCOPUS:85179886001
SN - 2352-7102
VL - 82
JO - Journal of Building Engineering
JF - Journal of Building Engineering
M1 - 108246
ER -