Existing kitchen make-up air system designs often overlook specific kitchen conditions and data-driven predictions. To address this, a dataset of 546 entries was created via experiments and CFD simulations. A novel XMS framework was developed, integrating XGBoost's feature interaction with an MLP's nonlinear mapping via stacking. This model optimizes the inlet area and airflow paths by predicting the temperature around the human body and the concentration of PM2.5 particles. The XMS algorithm reduced RMSE by 19.3% and improved R2 by 5.8% compared to polynomial regression, and outperformed a standalone MLP. For a 7.521 m2 kitchen with a 0.183 m & sup3;/s exhaust, an inlet area of 0.008-0.014 m2 and an airflow path of 5.9-6.9 m effectively maintained a body temperature above 290 K with low breathing-zone PM2.5. This study aids in selecting optimal ceiling make-up air systems for diverse kitchens.
The stack effect presents a significant challenge in high-rise building design, influencing thermal performance and fire safety. This study employs the airflow modeling software CONTAM to quantitatively assess the stack effect and pressure differentials in a building. Key variables examined include building height, HVAC operation strategies, elevator zoning configurations, regional climatic conditions, and the performance of air sealing interventions. The study further explores the impact of emerging multicar ropeless (MCR) elevator systems. Simulation results demonstrate the strong sensitivity of stack-induced airflow to vertical temperature differentials and total building height. HVAC pressurization strategies and transfer floor zoning significantly influence vertical air migration and energy demand profiles. Air leakage pathways, particularly those near ground-level lobbies, are shown to dominate infiltration patterns, with targeted sealing proving highly effective. Climatic variation across Chinese regions contributes to substantial differences in stack pressure gradients, underlining the need for climate-responsive designs. Preliminary analysis of MCR systems suggests potential advantages through decreased shaft surface area and enhanced air sealing strategies. These findings advocate for an integrated approach to tall building design, encompassing climate-specific strategies, precise elevator zoning, and strategic air barrier implementation to enhance supertall buildings' energy efficiency and environmental performance.
Make-up air (MUA) systems are critical for reducing indoor concentrations of cooking-generated particles (CGPs) and enhancing range hood capture efficiency. Inhibiting the diffusion of particles that escape range hood capture is essential for minimizing human exposure. This study proposes a novel ceiling make-up air system (CMAS), which delivers air from the ceiling of an adjacent room, to reduce indoor CGP concentrations. The effectiveness of the CMAS in eliminating CGPs was compared with that of natural make- up air (NMAS) from a window and a vent located in the cabinet under the stove (RMAS). Furthermore, the effects of the MUA delivery angle and the MUA-to-exhaust airflow ratio on CMAS performance were investigated. The results showed that CMAS performance was highly dependent on the specific combination of its angle and ratio. A CMAS configuration with a 0.75 MUA ratio and a 90 degrees angle provided the optimal CGP elimination. The proposed CMAS demonstrated significantly higher efficiency in both particle removal and attenuating particle diffusion compared to the NMAS and RMAS configurations. These findings offer a clear optimization strategy for indoor ventilation systems to alleviate the adverse health effects of CGPs.
Effective HVAC control is crucial for occupant health but often relies on inefficient iterative optimization. To address this, this study introduces a data-driven inversion framework integrating Global Sensitivity Analysis (GSA), Probability Density Sampling (PDS), and Convolutional Neural Networks (CNN) for direct HVAC control. Using Computational Fluid Dynamics (CFD) datasets, GSA identified sensitive locations to guide PDS in selecting sparse monitoring points for CNN training. This allows the model to predict optimal air supply parameters based on environmental targets. Results demonstrate that the PDS strategy significantly reduces data volume while ensuring high accuracy. The framework achieved Mean Relative Errors under 4.6% for supply velocity and angle, and 0.15% for temperature. Flow field reconstruction confirmed the model's precision. Consequently, this research provides an efficient approach for inverse air supply parameter determination, significantly advancing intelligent ventilation strategies in modern HVAC systems.
Indoor microorganisms have impacted human health through aerosol transmission and inhalation exposure. However, the appropriate purification technologies and practical effectiveness are still underexplored. This study evaluated the performance of different purification technologies for bioaerosols and the effects of operating thermal environments. The Aspergillus niger, was used as an artificial pollutant, and a combined air sampling and culturable method was employed to determine the concentration of airborne aerosols. 35 permutations of experimental conditions with temperature of 13/18/23/28/33 degrees C and relative humidity of 25%/35%/45%/55%/65%/75%/85% were designed. Comparisons of filtration, photocatalysis, ozone, and plasma showed that plasma exhibited higher disinfection rates, approximately 82.1 +/- 4.17%, which was 2.13 times that of natural decay. The relative humidity, rather than temperature, had a significant impact on plasma purification. As the relative humidity increased from 25% to 85% (increment of 10%), the disinfection rate first increased and then decreased, and the peak-value was found at 23 degrees C/55%. The work quantifies the effective purification technology for bioaerosols and reveals the influence of air temperature and humidity, which contributes to the development of testing standards when evaluating purification efficiency, and advances the practical applications to consider the indoor air pollutants and operational thermal environment conditions in buildings.
Current methods for assessing exposure to extreme heat and air pollution depend mostly on readings from regulatory monitoring stations. We hypothesize that this does not accurately represent the different conditions people encounter in various microenvironments. This study integrates distributed outdoor networks, indoor monitors, and wearable personal devices to characterize exposures to heat and fine particulate matter (PM2.5) experienced by residents of low-income public housing in Elizabeth, New Jersey. We apply exceedance-based metrics to measure both the intensity and duration of exposures across different participant groups (youth, seniors, staff working for the housing authority and researchers). The results show clear heterogeneity: Seniors face the highest PM2.5 exposures, and youth show widely varying and often higher thermal exposures. Comparisons of wearable devices and local monitors with regulatory monitors show congruent thermal conditions but high local variation in PM2.5 values. These findings highlight major discrepancies between existing design assumptions and actual living conditions. The study concludes that HVAC engineering efforts, urban design methods, and public health measures should be guided by personal monitoring data to tackle differences in risk. Future research should connect personal sensor data with health outcomes and consider the combined health effects of extreme heat and air pollution.
As an indispensable element of the heating, ventilation, and air conditioning (HVAC) system, the air handling unit (AHU) displays pronounced nonlinear attribute and two-directional dynamics, i.e., the time-wise dynamics within each running day and batch-wise dynamics between various days. Nevertheless, the current fault detection methodologies are insufficiently adept at tackling the AHU' s two-directional dynamics hidden in the nonlinear running data. To elevate the effectiveness of the fault detection for the AHU, a pioneering multiple batches differential analysis based kernel canonical variate analysis (MDBKCVA) strategy is developed in this paper, where three contributions are made. The first contribution is to perform the differential analysis on multiple normal batch datasets collected from various operating days to acquire the three-dimensional differential training dataset, which highlights the fault characteristics information and suppresses the dynamic changes between diverse batch datasets. Specifically, the multiple normal batch datasets are first non-repetitively paired up, then the differences between each pair of normal batch datasets are calculated to obtain the corresponding normal differential batch datasets. To reduce the computational complexity, some representative normal differential batch datasets are selected to assemble the three-dimensional differential training dataset. The second contribution is to employ the multiple batches analysis (MBA) technique to convert the three-dimensional normal differential training dataset into a normalized variable-wise unfolding differential matrix to further tackle both the batch-wise dynamics and variables' correlations. The third contribution is to integrate the kernel trick into the CVA to establish the KCVA monitoring model, for the purpose of sufficiently settling the nonlinearity and the time-wise dynamics. Thereafter, the latent variables are obtained, three monitoring indices are constructed to detect the AHU's faults. Finally, experiments and comparisons on the ASHRAE RP-1312 datasets are carried out to assess the suggested MDBKCVA' s fault detection capability.
The indoor thermal environment is important for the health, comfort, and performance of medical staff. However, most hospital thermal comfort studies focus on patients, while healthcare personnel and their diverse work conditions remain less studied. This study investigated the winter thermal environment in a hospital in Lanzhou, China, using field measurements and 382 valid questionnaires from staff in four departments: outpatient, ultrasound, surgical ward, and medical ward. Environmental parameters, including air temperature, relative humidity, air velocity, globe temperature, CO2 concentration, and sound pressure level, were measured. Thermal comfort was assessed using the Predicted Mean Vote (PMV) model and regression analysis. The results show that PMV systematically underestimated staff thermal sensation, indicating limited applicability in this context. Neutral temperatures estimated by the Griffiths method ranged from 22.7 degrees C to 23.1 degrees C. The findings suggest that department-specific thermal management may better meet staff thermal requirements in cold-region hospitals.
In this study, we test the performance of Guideline 36-compliant sequences of operation for the air system relative to sequences of operation that predate Guideline 36. For this study, the air system comprises air handling units (AHUs) that serve variable air volume terminal units (VAVs). We present a platform that can be used to study the impact on building operations of changing the control system in a building, while holding constant any other aspect of that building, such as equipment upgrades, internal load variations, and changes to weather. We use the Virtual Cybernetic Building Testbed (VCBT), which contains commercial BACnet-enabled controllers and an operator workstation, as well as the Intelligent Building Agents Laboratory (IBAL), which uses chillers to meet the loads in the AHU-VAV system and can generate repeatable weather conditions and zone loads. These facilities use a hardware-in-the-loop approach, which combines a simulation of a building with hardware. This is the first demonstration of the combined VCBT-IBAL platform, which provides a method to evaluate potential sequences of operation in hardware before deploying them in real buildings.
In the United States, residential energy consumption has surpassed that of the commercial sector, placing a growing financial burden on homeowners due to rising utility costs. While significant optimization efforts have been applied to commercial buildings, the residential sector has lagged in adopting advanced efficiency measures, partly due to limited system-level data. This study develops machine learning (ML) models to predict compressor power consumption and supply air temperature for a residential split-system air conditioner in cooling mode. Extensive laboratory testing captured system behavior under varying operating conditions, and 40 different ML models were evaluated against multiple linear regression (MLR) as a baseline. XGBoost and a voting regressor achieved the highest accuracy across all investigated error metrics, while feature importance analysis identified outdoor air temperature and inlet air enthalpy as dominant predictors. While results are based on a single system tested under controlled conditions, future work should validate models in real-world settings and across diverse HVAC types. These models can support performance optimization, predictive maintenance, and regulatory compliance, enabling more energy-efficient residential HVAC operation.
Building ventilation controls often combine demand control ventilation (DCV), which reduces ventilation during low occupancy, with economizer cooling, which increases outdoor air when conditions allow free cooling. However, these strategies overlook outdoor air pollution, and economizers can worsen indoor air quality (IAQ) when outdoor fine particulate matter (PM2.5) is high. This study evaluates the real-word operation of the IAQ-Energy Controller, a rule-based controller that enhances Economizer + DCV logic by limiting outdoor air and disabling the economizer when outdoor PM2.5 is elevated. To compensate, the system modulates an internet-connected portable air cleaner to meet ASHRAE Standard 241 for infectious aerosol control. Implemented in a repeated-measures crossover field study across two classrooms in California's Central Valley, the controller consistently reduced carbon dioxide concentrations by 16% to 18% relative to fixed-rate ventilation, maintained or improved equivalent air changes per hour for infectious aerosol removal, and delivered comparable thermal comfort without increasing energy use. The infectious aerosol control benefits of the additional outside air provided by the IAQ-Energy controller increase as filtration efficiency of recirculated air decreases, which is important for applications with lower efficiency filters. The IAQ-Energy controller represents a low-cost, scalable retrofit solution for improving IAQ and resilience in school environments.
Local, on-tube condensation heat transfer coefficients of R134a alternatives, R450A and R513A, were measured on a single, horizontal, smooth 15.875-mm-diameter stainless-steel tube at saturation temperatures of 35 degrees C and 40 degrees C in a quiescent vapor space for a range of subcoolings (i.e., 1.7 degrees C-7.2 degrees C) and heat fluxes (i.e., 5-16 kW/m2). Gravity-driven flows were visualized through site glasses and droplet-mode flow was observed. Experiments were validated using Nusselt's correlation for on-tube condensation. Local heat fluxes were used to compute local on-tube condensation heat transfer coefficients and the wall temperature was directly measured. Data were presented for two radial positions (i.e., angles beta = 11 degrees and beta = 109 degrees from the vertical). For both refrigerants, local heat transfer coefficients at beta = 11 degrees were higher than at beta = 109 degrees due to the thinner condensed film. At the same subcooling and at a saturation temperature of 35 degrees C and 40 degrees C, local heat transfer coefficients for R513A were higher than R450A. Local heat transfer coefficients were also compared between new (i.e., 3-6 months) and stored (i.e., 3-year old) R513A and R450A, and there were minimal differences. A gas chromatography analysis, and analysis using the vapor liquid equilibrium assumption, showed that there were modest composition changes for either one or both stored refrigerant components.
