Machine learning–assisted prediction of heat fluxes through thermally anisotropic building envelopes

Zhenglai Shen; Som Shrestha; Daniel Howard; Tianli Feng; Diana Hun; Buxin She

doi:10.1016/j.buildenv.2023.110157

Machine learning–assisted prediction of heat fluxes through thermally anisotropic building envelopes

Zhenglai Shen, Som Shrestha, Daniel Howard, Tianli Feng, Diana Hun, Buxin She

Building Envelope Materials Research

Research output: Contribution to journal › Article › peer-review

14 Scopus citations

Abstract

Thermally anisotropic building envelope (TABE) is a novel active building envelope that can save energy use to maintain thermal comfort in buildings by redirecting heat and coolness from building envelopes to thermal loops. Finite element models (FEMs) can be used to compute the heat fluxes through TABEs, but the high computational cost of finite element simulations has prevented parametric studies and design optimizations. This paper proposes a domain knowledge–informed, finite element–based machine learning framework to reduce the computation cost for the energy management of buildings installed with TABE that uses a ground thermal loop. First, the training heat flux data set was generated by FEM simulations with different thermal loop schedules. Then, both shallow learning models (i.e., multivariate linear regression and eXtreme Gradient Boost, or XGBoost) and a deep learning model (i.e., deep neural network, or DNN) were trained to predict the heat fluxes. Domain knowledge was used for data preprocessing and feature selection. Finally, the suitability of the selected machine learning model was tested under different thermal loop schedules. The case study results showed that: (1) XGBoost can be as accurate as DNN (coefficient of determination equal to 0.81) with much less training time; (2) the annual energy cost savings for different thermal loop schedules obtained by the XGBoost-predicted and FEM-calculated heat fluxes are consistent, having a difference of only 4%; and (3) XGBoost can reduce the computation time for the annual energy analysis of the case study building with a given thermal loop schedule from around 12 h by using FEM to less than 1 min.

Original language	English
Article number	110157
Journal	Building and Environment
Volume	234
DOIs	https://doi.org/10.1016/j.buildenv.2023.110157
State	Published - Apr 15 2023

Funding

This research was supported by the DOE's Office of Energy Efficiency and Renewable Energy , Building Technologies Office under Contract No. DE-AC05-00OR22725 with UT-Battelle LLC and used resources at the Building Technologies and Research Integration Center, a DOE-EERE User Facility at Oak Ridge National Laboratory. ML is one of the most used data-driven methods that combines computer science and statistics, and it serves as the core of artificial intelligence and data science [27]. Unlike the physics-based model, the ML model makes a prediction without knowing the detailed physics. The ML model is also known as the black-box model [28]. ML has been widely used in building energy–related studies to predict a building's thermal load [29–31], energy consumption [32–34], wall heat flux [35], heat loss coefficient [36], and other related factors. Based on the model's structure, ML models can be categorized into shallow learning models and deep learning models. Shallow learning models learn from data described by predefined features and usually consist of very few layers of composition. Examples include Multivariate Linear Regression (MaLR), support vector machine, decision tree, random forest, eXtreme Gradient Boost (XGBoost), and Artificial Neural Network (ANN) with one hidden layer. ANN has a strong fitting capability owing to its thousands of neurons and nonlinear activation functions, which may represent a wide variety of functions when given appropriate weights and biases [37].The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Som Shrestha reports financial support was provided by US Department of Energy. Som Shrestha has patent Thermally Anisotropic Composites for Thermal Management in Building Environments pending to UT-BATTELLE, LLCThis research was supported by the DOE's Office of Energy Efficiency and Renewable Energy, Building Technologies Office under Contract No. DE-AC05-00OR22725 with UT-Battelle LLC and used resources at the Building Technologies and Research Integration Center, a DOE-EERE User Facility at Oak Ridge National Laboratory. Notice: This manuscript has been authored by UT-Battelle, LLC, under contract DE-AC05-00OR22725 with the US Department of Energy (DOE). The US government retains and the publisher, by accepting the article for publication, acknowledges that the US government retains a nonexclusive, paid-up, irrevocable, worldwide license to publish or reproduce the published form of this manuscript, or allow others to do so, for US government purposes. DOE will provide public access to these results of federally sponsored research in accordance with the DOE Public Access Plan ( http://energy.gov/downloads/doe-public-access-plan ).

Funders	Funder number
Artificial Neural Network
DOE-EERE
MaLR
U.S. Department of Energy
Oak Ridge National Laboratory
Building Technologies Office	DE-AC05-00OR22725
UT-Battelle

Keywords

Building energy management
Heat flux prediction
Machine learning
Thermally anisotropic building envelope (TABE)

Access to Document

10.1016/j.buildenv.2023.110157

Cite this

@article{29652af0b905400ca54b4efce5994d74,

title = "Machine learning–assisted prediction of heat fluxes through thermally anisotropic building envelopes",

abstract = "Thermally anisotropic building envelope (TABE) is a novel active building envelope that can save energy use to maintain thermal comfort in buildings by redirecting heat and coolness from building envelopes to thermal loops. Finite element models (FEMs) can be used to compute the heat fluxes through TABEs, but the high computational cost of finite element simulations has prevented parametric studies and design optimizations. This paper proposes a domain knowledge–informed, finite element–based machine learning framework to reduce the computation cost for the energy management of buildings installed with TABE that uses a ground thermal loop. First, the training heat flux data set was generated by FEM simulations with different thermal loop schedules. Then, both shallow learning models (i.e., multivariate linear regression and eXtreme Gradient Boost, or XGBoost) and a deep learning model (i.e., deep neural network, or DNN) were trained to predict the heat fluxes. Domain knowledge was used for data preprocessing and feature selection. Finally, the suitability of the selected machine learning model was tested under different thermal loop schedules. The case study results showed that: (1) XGBoost can be as accurate as DNN (coefficient of determination equal to 0.81) with much less training time; (2) the annual energy cost savings for different thermal loop schedules obtained by the XGBoost-predicted and FEM-calculated heat fluxes are consistent, having a difference of only 4%; and (3) XGBoost can reduce the computation time for the annual energy analysis of the case study building with a given thermal loop schedule from around 12 h by using FEM to less than 1 min.",

keywords = "Building energy management, Heat flux prediction, Machine learning, Thermally anisotropic building envelope (TABE)",

author = "Zhenglai Shen and Som Shrestha and Daniel Howard and Tianli Feng and Diana Hun and Buxin She",

note = "Publisher Copyright: {\textcopyright} 2023 Elsevier Ltd",

year = "2023",

month = apr,

day = "15",

doi = "10.1016/j.buildenv.2023.110157",

language = "English",

volume = "234",

journal = "Building and Environment",

issn = "0360-1323",

publisher = "Elsevier",

}

TY - JOUR

T1 - Machine learning–assisted prediction of heat fluxes through thermally anisotropic building envelopes

AU - Shen, Zhenglai

AU - Shrestha, Som

AU - Howard, Daniel

AU - Feng, Tianli

AU - Hun, Diana

AU - She, Buxin

PY - 2023/4/15

Y1 - 2023/4/15

N2 - Thermally anisotropic building envelope (TABE) is a novel active building envelope that can save energy use to maintain thermal comfort in buildings by redirecting heat and coolness from building envelopes to thermal loops. Finite element models (FEMs) can be used to compute the heat fluxes through TABEs, but the high computational cost of finite element simulations has prevented parametric studies and design optimizations. This paper proposes a domain knowledge–informed, finite element–based machine learning framework to reduce the computation cost for the energy management of buildings installed with TABE that uses a ground thermal loop. First, the training heat flux data set was generated by FEM simulations with different thermal loop schedules. Then, both shallow learning models (i.e., multivariate linear regression and eXtreme Gradient Boost, or XGBoost) and a deep learning model (i.e., deep neural network, or DNN) were trained to predict the heat fluxes. Domain knowledge was used for data preprocessing and feature selection. Finally, the suitability of the selected machine learning model was tested under different thermal loop schedules. The case study results showed that: (1) XGBoost can be as accurate as DNN (coefficient of determination equal to 0.81) with much less training time; (2) the annual energy cost savings for different thermal loop schedules obtained by the XGBoost-predicted and FEM-calculated heat fluxes are consistent, having a difference of only 4%; and (3) XGBoost can reduce the computation time for the annual energy analysis of the case study building with a given thermal loop schedule from around 12 h by using FEM to less than 1 min.

AB - Thermally anisotropic building envelope (TABE) is a novel active building envelope that can save energy use to maintain thermal comfort in buildings by redirecting heat and coolness from building envelopes to thermal loops. Finite element models (FEMs) can be used to compute the heat fluxes through TABEs, but the high computational cost of finite element simulations has prevented parametric studies and design optimizations. This paper proposes a domain knowledge–informed, finite element–based machine learning framework to reduce the computation cost for the energy management of buildings installed with TABE that uses a ground thermal loop. First, the training heat flux data set was generated by FEM simulations with different thermal loop schedules. Then, both shallow learning models (i.e., multivariate linear regression and eXtreme Gradient Boost, or XGBoost) and a deep learning model (i.e., deep neural network, or DNN) were trained to predict the heat fluxes. Domain knowledge was used for data preprocessing and feature selection. Finally, the suitability of the selected machine learning model was tested under different thermal loop schedules. The case study results showed that: (1) XGBoost can be as accurate as DNN (coefficient of determination equal to 0.81) with much less training time; (2) the annual energy cost savings for different thermal loop schedules obtained by the XGBoost-predicted and FEM-calculated heat fluxes are consistent, having a difference of only 4%; and (3) XGBoost can reduce the computation time for the annual energy analysis of the case study building with a given thermal loop schedule from around 12 h by using FEM to less than 1 min.

KW - Building energy management

KW - Heat flux prediction

KW - Machine learning

KW - Thermally anisotropic building envelope (TABE)

UR - http://www.scopus.com/inward/record.url?scp=85150814519&partnerID=8YFLogxK

U2 - 10.1016/j.buildenv.2023.110157

DO - 10.1016/j.buildenv.2023.110157

M3 - Article

AN - SCOPUS:85150814519

SN - 0360-1323

VL - 234

JO - Building and Environment

JF - Building and Environment

M1 - 110157

ER -

Machine learning–assisted prediction of heat fluxes through thermally anisotropic building envelopes

Abstract

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this