The University of Tokyo · Environmental Science
Professor Chenghao Wei's research lab specializes in computational fluid dynamics (CFD), energy efficiency in commercial buildings, and thermal comfort optimization, with a strong focus on supermarkets and grocery stores as key energy-intensive environments. The lab develops advanced machine learning techniques—such as deep neural networks and Bayesian networks—combined with proper orthogonal decomposition (POD) to accelerate unsteady CFD simulations and improve predictive accuracy with limited training data. Research also emphasizes practical energy-saving strategies, including optimized display case design and ventilation control, to reduce energy consumption while enhancing indoor environmental quality.
Figures are computed from collected data and may differ slightly.
Abstract Computational fluid dynamics (CFD) is widely used to predict the indoor thermal environment; however, large time cost represents a significant disadvantage. Several deep learning approaches have been introduced to reduce prediction time in steady‐state predictions, though their feasibility under unsteady ones has yet to be investigated. Considering the flexibility of the multilayer perceptron (MLP) input–output format, this study compared the performance of two MLP input–output formats,
The purpose of this study is evaluating energy consumption and the indoor thermal environment of grocery stores, which using frozen and refrigerated display cases, by taking temperature distribution into consideration. In this report, store CFD models which can evaluate the indoor temperature distribution were adopted to study the influence, made by temperature or ventilation flux altering, on energy consumption of grocery stores. The results indicate that the mixture of indoor air, no matter af
In this study, to predict unsteady temperature distributions, POD-DNN was utilized, where DNN was trained to predicted coefficients of POMs. Two strategies, flatten POD-DNN and nested POD-DNN were compared. The flatten POD-DNN provided high accuracy if training data is sufficient, but otherwise very inaccurate. The nested POD-DNN roughly predicted the development of temperature fields even training data was small. The results showed their different sensitivities to the training data size.
Supermarkets and grocery stores represent an important energy-intensive role in the commercial sector. In Japan, grocery stores occupy 8% of energy consumption in the commercial sector. Refrigerated and frozen display cases are widely used in grocery stores and those occupation of energy consumption in a store range from 50% to 67%. In addition, plenty of research illustrated that alterations in humidity and temperature affect the cooling load of display cases. Also, cold air leaking from displa
In this report, some energy-saving and thermal comfort improving strategies adopted in a grocery store were estimated by using the store environment analysis CFD model. As results, by replacing the frozen multi-deck display cases with frozen reach-in cases, total energy consumption of cases and air conditioners decreased to 3/4 both in summer and winter, meanwhile it led to a more comfortable thermal environment. In winter, exhausting from the lower part of display cases resulted in 3% decreasin
For Bayesian network structure learning with continuous data, traditional methods typically require data discretization or assume that the data follows a Gaussian distribution. However, the processing method can result in information loss, and real-world data often deviates from the Gaussian assumption, leading to biased outcomes. To address these issues, we propose using the k-nearest neighbor (k-NN) algorithm to estimate mutual information and conditional mutual information for Bayesian networ
Open papers in the app to read, cite, and organize with AI.
