Skip to main content

Performance Evaluation of Gray-box and Machine Learning Models of a Thermal Energy Storage System with Active Insulation

by Borui Cui, Jin Dong, Piljae Im, Sungkyun Jung, Melissa V Lapsa
Publication Type
Conference Paper
Book Title
8th Thermal and Fluids Engineering Conference (TFEC)
Publication Date
Page Numbers
609 to 618
Publisher Location
Connecticut, United States of America
Conference Name
8th Thermal and Fluids Engineering Conference (TFEC)
Conference Location
College Park, Maryland, United States of America
Conference Sponsor
University of Maryland
Conference Date

An interior partition wall integrated with active thermal storage and a dynamic insulation system was built and then installed in an office building in Oak Ridge, Tennessee, TN. This smart wall, termed the Empower Wall, was equipped with embedded pipes in the building envelope core component and an additional pipe network enclosing rigid insulation to switch on and off the active insulation dynamically. The performance of the wall's contribution to cooling load reduction under different parameters has been investigated in previous publications. Aiming to be deployed into model predictive control and other optimization methods, simplified and reliable models for the developed wall and the room accommodating it are required. They are needed to characterize the properties and thermal response of both Empower Wall and building envelope, which form an essential component for accurate indoor temperature or cooling/heating demand prediction. In this study, simplified gray-box and regression models as well as machine learning model were developed and the performance of them were compared and analyzed.