Introduction

This semester studying how buildings can operate more efficiently to better support occupants and promote their health culminates in a workshop where seven groups are sharing their findings on new techniques to make buildings smarter. The workshop is organized into three sessions: Environment and Perception, HVAC Control, and Understanding Sleep, Productivity, and Distractions, and highlights new research on the future of smart and healthy buildings.

Time and Location

  • Date: Tuesday, December 5, 2023
  • Time: 9:00AM - 10:45AM
  • Location: Rice 340

Schedule

Time Paper
  Session 1: Environment and Perception
9:00-9:15 Enhancing Perceived Thermal Comfort: Exploring the Impact of Lighting Factors
9:15-9:30 The Effects of Aromatic VOCs on Psychological and Physiological Stress in an Office
  Session 2: HVAC Control
9:30-9:45 Enhancing Indoor Air Quality And HVAC Control Through Noise-Based HVAC Optimization
9:45-10:00 Evaluating HVAC Zone Efficiency by Resolution: Creating Smart and Healthy Buildings
  Session 3: Understanding Sleep, Productivity, and Distractions
10:00-10:15 Impact of Daily Activities and Screen Time on Sleep Quality: A Cross-Sectional Study
10:15-10:30 Predicting Productivity Using Fitbit and Environmental Data
10:30-10:45 Analyzing & Expanding The Accuracy of AI Based Image Models on Distraction Detection

Papers

Enhancing Perceived Thermal Comfort: Exploring the Impact of Lighting Factors

Abstract: This study explores the impact of ambient lighting on occupants’ perceived thermal comfort in response to the environmental challenges posed by traditional temperature control systems. Three lighting scenarios are tested: control, warm light, and cool light. Hypotheses suggest that warm lighting may enhance comfort, while cool lighting may have the opposite effect. Thirteen participants underwent the experiment, providing data on environmental factors, true thermal comfort, and perceived thermal comfort. The study showed no discernible relationship between perceived thermal comfort, lighting factors, previous activities, or quantity of sleep. With that said the inconclusive results of this study aim to inform future researchers of best practices in experimental design and address strategies for sustainable and occupant-friendly design in a changing climate. Understanding the role of lighting in perceived thermal comfort informs future building.

Authors: Nora LeVasseur, Naeem Patel, Prashanth Wagle, Gabriel Silliman

The Effects of Aromatic VOCs on Psychological and Physiological Stress in an Office

Abstract: Robust evidence demonstrates the benefits of exposure to nature on mental health. Biophilic design seeks to translate the benefits of nature to architectural design, and has been shown to relieve stress and improve perception of environments. Aromatherapy is a widely known form of alternative medicine. Though there is no strong evidence that links aromatherapy with curing disease, aromatherapy can significantly impact the nervous system and bring a sense of naturalness to an interior environment. Not many studies have explored the impact of aromatherapy on indoor air quality and stress. By introducing scents, the amount of volatile organic compounds (VOCs) in the environment increase. Often, higher VOCs are an indicator of bad air quality. This study attempts to see if lavender essential oil can impact stress recovery despite lowering the air quality. In this study, participants’ heart rate and mood were tracked as they were tasked with completing an interview question and a Stroop test in both a ‘control’ and lavender ‘enriched’ office. The main objective is to see whether participants’ heart rate and perceived mental health are aided by this type of olfactory stimulation.

Authors: Joseph Loggi, Dhrumil Joshi, Hamza Khalid, Katherine McCarthy, Julianna Mollica

Enhancing Indoor Air Quality And HVAC Control Through Noise-Based HVAC Optimization

Abstract: This study explores the innovative integration of ma- chine learning with HVAC systems to optimize indoor air quality (IAQ) and energy efficiency in urban environments. Focusing on the Link Lab at the University of Virginia, we examine the correlation between ambient noise levels and CO2 concentrations, hypothesizing that noise can be a predictive metric for IAQ. Utilizing a dataset comprising noise, CO2 levels, and room features, our approach employs recurrent neural network (RNN) models to predict CO2 concentrations, facilitating proactive HVAC adjustments. This research contributes to the emerging field of smart building management, offering a novel approach to IAQ optimization using ambient noise as a predictive tool, thereby enhancing both occupant comfort and environmental sustainability.

Authors: Shubhankar Poundrik, Fitzgerald Marcelin, Muchun Liu

Evaluating HVAC Zone Efficiency by Resolution: Creating Smart and Healthy Buildings

Abstract: This paper explores the concept of ‘resolution’ in Heating, Ventilation, and Air Conditioning (HVAC) systems, specifically examining its impact on energy efficiency and operational costs in the context of HVAC zones. The study focuses on the Link Lab at the University of Virginia, using a case study of three adjacent office spaces within the same HVAC zone to assess the potential benefits of higher-resolution zone control. We examine the current state of HVAC operations, including energy consumption and control strategies, and propose an innovative model for zoning division. Our evaluation includes a comparison of energy usage between the current system and a hypothetical high-resolution zone control scheme. The findings suggest that higher resolution in HVAC systems can lead to significant energy savings and cost reductions, although implementation challenges and trade-offs exist. The study concludes with a discussion of the feasibility and implications of implementing high-resolution HVAC systems in similar environments.

Authors: Stuart Paine, Adam Khan, Joseph Remines

Impact of Daily Activities and Screen Time on Sleep Quality: A Cross-Sectional Study

Abstract: Sleep quality has a significant impact on human health and behavior. It is influenced by various factors, including air quality and room temperature. This cross-sectional study aimed to examine the impact of daily activities and screen time on sleep quality. The study analyzed data from 13 participants over three months and included self-reported sleep quality, screen time, and steps. The study discovered that individuals who spent less time using screens generally exhibited better sleep quality. However, no significant correlation was observed between step count and sleep quality. A heat map of the correlation matrix depicted low negative correlations between screen time, step count, and sleep quality. Linear regression analysis was utilized to gain further insight into the influence of these variables on sleep quality. This study offers valuable insights into promoting better sleep quality by adjusting screen time and step counts. It contributes to a better understanding of the factors that impact sleep quality and offers guidance for interventions to improve sleep habits.

Authors: JiHo Lee, Zian Jin, Vinay Vangala, Ajay Sanjeevan, Anushruti Huilgol

Predicting Productivity Using Fitbit and Environmental Data

Abstract: Recent research has underscored the correlation between the work environment and employee performance, particularly emphasizing the significance of indoor environmental quality components such as thermal comfort and acoustic conditions. These factors profoundly influence an individual’s focus and effectiveness in the workplace. However, optimizing indoor spaces to enhance productivity necessitates a nuanced comprehension of the specific environmental factors directly shaping occupants’ experiences.

To gain a deeper insight into the quantitative relationships between diverse environmental factors and productivity levels, our study leverages existing data involving 13 students and faculty members from the University of Virginia (UVA). Our predictive models successfully pinpoint primary external factors that impact productivity, encompassing environmental elements like air quality and individual aspects such as sleep quality. Consequently, our study can forecast productivity levels and elucidate their combined impact on overall productivity.

The findings from our study reveal several critical factors affecting productivity. First, the day of the week emerges as a significant predictor, indicating lower productivity during weekends and heightened productivity on Thursdays and Fridays. Second, air quality measurements are potentially important predictors, with higher values generally associated with reduced productivity (except for VOCs, which warrants further exploration). Lastly, our study highlights that deeper sleep positively influences productivity, while sleep duration does not notably impact it.

Authors: Mitchell Whalen, Emily Branch, Yuanyuan Zhao, Ishaan Sanghvi

Analyzing & Expanding The Accuracy of AI Based Image Models on Distraction Detection

Abstract: In our rapidly evolving digital era, where technological progress constantly reshapes the landscape of productivity, the task of maintaining focus and minimizing distractions has emerged as a crucial challenge. Individuals navigating through work and study encounter a wide range of potential distractions, these may be leisure websites, mobile devices or even humans beside them. To increase and track productivity, the exploration of novel solutions to comprehend, analyze, and mitigate distractions has become more imperative than ever. An innovative approach to address this challenge involves leveraging video analysis to track and interpret distractions in real-time. The application of video analysis has transcended its conventional use in surveillance, evolving into a potent tool for studying human behavior and interactions. Through the implementation of computer vision techniques, video analysis programs now exist tracking distractions mainly through gaze tracking and head movement of the person recorded by facecam.

Authors: Chengyuan Cai, Lance Tecson, Kofi Darfour

Data-Driven Energy Estimation for Indoor Solar-Powered Energy Harvesting Sensors

Github Paper Authors: Viswajith Govinda Rajan, Zoraiz Qureshi, Akash Nair, Zichuan Guo Abstract: Energy generation fluctuation leads to uncertainty in the estimation of energy availability which can influence the design of in-door energy harvesting Internet of Things (IoT) devices. Lux sensors may also not be available every time for this estimation, and their estimation is very sensitive to placement. We explore methods for robust data-driven estimation of energy availability for indoor energy harvesting sensors, focusing specifically on solar-powered sensors from different sensors as sources such as motion and acoustic noise sensor data.

A real-time indoor environmental status dashboard

Github Paper Authors: Xinyue Fan, Blake Wang, Anne Zhang, Francis Becker Abstract: We created a web interface that tracks real-time sensor data from select spaces in the Link Lab to generate text prompts for Craiyon, a free and accessible AI text to image generator. We will specifically focus on CO2, temperature and illumination data provided by the linklab sensors which will reflect on the indoor comfort level of the office space. The data values for each category will be used to determine text prompts that generates a coherent image depicting brightness, occupancy and temperature and displayed through an interactive real time dashboard for visualization.

A Cost-Effective Non-Invasive Method for Accurate Occupancy Detection in Conference Rooms Using Environmental Quality Sensors

Github Paper Authors: Kane Aldrich, Katheryn Flynn, Hamidreza Nabaei, Emmett Rice Abstract: The increasing cost of energy and a greater knowledge of the impact of the environment on human health and pro- ductivity has led to an increased demand for smart solutions and dynamic control to the Heating, Ventilation, and Air Conditioning (HVAC) of buildings. In the context of smart response, the knowledge of occupants in a room allows the HVAC system to respond to the natural degradation of a closed environment when populated. This paper presents an approach using Indoor Air Quality (IAQ) sensors to create a model for the number of occupants in a room. 86% accu- racy in exact occupant number within a conference room is achieved when using a singular 𝐶𝑂2 sensor. The method pre- sented also reduces the cost of sensing by using IAQ sensors rather than intrusive cameras, or high cost motion detection.

Room Occupant Identification using Machine Learning

Github Paper Authors: Yue Zhu, Xavier Castillo-Vieira, Runnan Zhou, Jiahao Shen, Shuhao Dong Abstract: The room reservation process is being equipped with automation tools in recent years for faster and more convenient service. However, existing systems are not smart enough to verify whether a room is occupied by the group that made the reservation. In this paper, we present a machine-learning-based automation occupancy detection tool to tell whether a periodically reserved room is actually used by the intended applicant or group. We trained the classifier using three machine learning models(SVM, KNN and Bayesian) separately and evaluated their performances on the real-world data set. KNN achieved the best accuracy of 63.5\% among the tested models, which indicates the feasibility of classifying a group’s identity based on spatial sensor readings to a reasonable accuracy level.

Measuring Accuracy in Occupant Recall of Environmental Comfort Indicators

Github link Paper link Authors: Tao Jiang, Haoqian Li, Richard Yu, Erzhen Hu, Natalie Ownby Abstract: Many existing studies regarding individual comfort preferences in indoor environments rely on occupants’ recall of environmental trends. Indeed, their reported comfort levels can be greatly affected by their capacity for recall. However, the existing literature has a blind spot when measuring the accuracy of individual perception. In this paper. We conducted a survey of thirteen occupants in the University of Virginia’s Link Lab, which is a climate-controlled smart building which measures environmental quality via a network of sensors. In the survey, we asked about the general trends in environmental comfort indicators (Humidity, Temperature, Noise levels, Light levels, and Air Quality) that they could recall in the Link Lab. Then, we compared their perceptions to actual results pulled from the Link Lab sensors during a similar timeframe over one week. Our results indicated that there did exist some discrepancies when it came to individual opinion and observed trends.