EVALUATION OF OCCUPANCY PATTERNS IN RESIDENTIAL BUILDING OF THE UNITED STATES AND THE IMPACT OF PANDEMIC ON THE PROFILES By Debrudra Mitra A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Civil Engineering – Doctor of Philosophy 2022 ABSTRACT The energy performance of a residential building is highly dependent on the occupant's presence or non-presence in a building and their interactions with energy-consuming appliances. Typical occupancy schedules for residential buildings must be defined for applications such as building energy modeling as well as for assessing energy savings associated with the use of occupancy sensing technologies and occupancy-dependent controls. Currently, commonly used simulation programs assume a typical occupancy schedule, however, there is significant opportunity for improvement to these schedules as this is generally based on engineering judgement. This research uses multiple years of the American Time Use Survey (ATUS) data to develop typical occupancy schedules for a range of household types and based on different occupant characteristics and compared to currently utilized residential occupancy schedules. The impact of COVID-19 pandemic on the occupancy schedules is also analyzed in this study. The results of this research work towards improved occupancy schedule development can benefit both industry professionals and researchers. This thesis is dedicated to my lovely nieces Elina and Ahana iii ACKNOWLEDGEMENTS I would like to thank many people who helped me along the way on this journey. I would like to take a moment to appreciate them without whom this work cannot be complete. Firstly, I am honored to express my sincere gratitude to my advisor Dr. Kristen Cetin. I am privileged to find Dr. Cetin as my teacher and mentor. To me, she acts as mentor, motivator, and a role model to follow in my future. She was very supportive, a great motivator and provided continuous source of encouragement to explore new ideas and improve the work quality. She was open to listen to new ideas, patient to hear my challenges and always provided constructive feedback to help me improve as a researcher as well as human being. I could not imagine having any other advisor for my PhD work. Along with her, I would also like to thank the rest of the members of my PhD committee, Dr. Dong Zhao, Dr. Menrnaz Ghamami and Dr. Sharlissa Moore for providing valuable suggestions and insightful feedbacks to improve the overall study. I am also thankful to my colleagues and friends, Yiyi Chu, Soham Vanage and Hao Dong for their continuous support, valuable suggestions and taking extra step to help in different scenarios. I also like to thank my friends like Ritam Ganguly, Oyendrila Dobe, Arna Ganguly and many others to make the PhD journey an enjoying and fun time. Finally, I want to thank my parents who have supported me in every up and down of my life and have encouraged me all the times without expecting anything in return. iv 1. TABLE OF CONTENTS 1. CHAPTER 1 – INTRODUCTION ......................................................................................... 1 1. Research needs and hypotheses .............................................................................................. 1 2. Research Objectives .............................................................................................................. 10 3. Organization.......................................................................................................................... 13 4. Research Organization .......................................................................................................... 13 REFERENCES ......................................................................................................................... 15 2. CHAPTER 2 - TYPICAL OCCUPANCY PROFILES AND BEHAVIORS IN RESIDENTIAL BUILDINGS IN THE UNITED STATES ........................................................ 23 1. Abstract ................................................................................................................................. 23 2. Keywords .............................................................................................................................. 23 3. Introduction ........................................................................................................................... 24 4. Occupancy Data .................................................................................................................... 27 5. Methodology ......................................................................................................................... 29 6. Results and Discussion ......................................................................................................... 31 7. Conclusions ........................................................................................................................... 44 8. Acknowledgements ............................................................................................................... 46 REFERENCES ......................................................................................................................... 47 APPENDIX ............................................................................................................................... 51 3. CHAPTER 3 – CLUSTER ANALYSIS OF OCCUPANCY SCHEDULES IN RESIDENTIAL BUILDINGS IN THE UNITED STATES ........................................................ 58 1. Abstract ................................................................................................................................. 58 2. Introduction ........................................................................................................................... 59 3. Datasets ................................................................................................................................. 63 4. Methodology ......................................................................................................................... 65 5. Results ................................................................................................................................... 67 6. Conclusions ........................................................................................................................... 81 7. Acknowledgements ............................................................................................................... 83 REFERENCES ......................................................................................................................... 84 APPENDIX ............................................................................................................................... 88 4. CHAPTER 4 - VARIATION IN RESIDENTIAL OCCUPANCY PROFILES IN THE UNITED STATES BY HOUSEHOLD INCOME LEVEL AND CHARACTERISTICS ........ 103 1. Abstract ............................................................................................................................... 103 2. Introduction ......................................................................................................................... 103 3. Datasets ............................................................................................................................... 107 4. Methodology ....................................................................................................................... 109 5. Results and Discussion ....................................................................................................... 112 6. Conclusions ......................................................................................................................... 129 7. Acknowledgements ............................................................................................................. 131 REFERENCES ....................................................................................................................... 132 v 5. CHAPTER 5. COVID-19 IMPACTS ON RESIDENTIAL OCCUPANCY SCHEDULES AND ACTIVITIES IN U.S. HOMES IN 2020 USING ATUS.................................................. 140 1. Abstract ............................................................................................................................... 140 2. Keywords ............................................................................................................................ 141 3. Introduction ......................................................................................................................... 141 4. Dataset................................................................................................................................. 144 5. Methodology ....................................................................................................................... 145 6. Results and Discussion ....................................................................................................... 147 7. Conclusions ......................................................................................................................... 161 8. Acknowledgements ............................................................................................................. 164 REFERENCES ....................................................................................................................... 165 APPENDIX ............................................................................................................................. 171 6. CHAPTER 6 – CONCLUSION AND FUTURE WORK .................................................. 179 1. Conclusions ......................................................................................................................... 179 2. Limitation............................................................................................................................ 183 3. Research contribution ......................................................................................................... 184 4. Future work ......................................................................................................................... 185 7. CHAPTER 7 – PUBLICATIONS ...................................................................................... 186 1. Journal Publications ............................................................................................................ 186 2. Book Chapter ...................................................................................................................... 186 3. Conference Proceedings...................................................................................................... 186 4. Others .................................................................................................................................. 187 vi 1. CHAPTER 1 – INTRODUCTION 1. Research needs and hypotheses The building sector is a major energy consumer throughout the world. In the United States in particular, the percentage of energy consumed by the building sector has increased significantly compared to the other sectors in the past 70 years [1] (Figure 1.a). In addition, electricity consumption by the building sector significantly dominates the other sectors (Figure 1.b). The U.S. Energy Outlook (EIA 2020) reports that the purchased electricity used by the residential and commercial building sectors is projected to increase an average of 0.6% and 0.8% per year through 2050 [2]. Energy consumption from the building sector accounts for around 30% of all greenhouse gas (GHG) emissions worldwide [3]. In the United States in 2018, 12.3% of the overall GHG emissions from direct energy consumed originates from the building sector [4]. Given the need to curb GHG emissions to reduce the impacts of climate change (IPCC 2014), it is important to develop and implement strategies to reduce the consumption of the U.S. building stock [5]. 50 (a) Energy consumption 40 by sector (%) 30 20 10 194 19 9 5 19 2 5 19 5 5 19 8 6 19 1 6 19 4 6 19 7 7 19 0 7 19 3 7 19 6 7 19 9 8 19 2 8 19 5 8 19 8 9 19 1 9 19 4 9 20 7 0 20 0 0 20 3 0 20 6 0 20 9 1 20 2 1 20 5 18 Year 80 Building sector Industrial Sector Transportation Sector (b) Electricity consumption 70 60 50 40 by sector (%) 30 20 10 0 19 4 19 9 5 19 2 5 19 5 5 19 8 6 19 1 6 19 4 6 19 7 7 19 0 7 19 3 7 19 6 7 19 9 8 19 2 8 19 5 8 19 8 9 19 1 9 19 4 9 20 7 0 20 0 0 20 3 0 20 6 0 20 9 1 20 2 1 20 5 18 Year Building sector Industrial Sector Transportation Sector Figure 1-1. Annual percentage variation in (a) energy and (b) electricity consumption for different sectors (U.S. EIA, [1]) 1 As shown in Figure 2, the residential building sector consistently consumes more energy compared to commercial buildings [2]. Although the difference between the residential and commercial sector has reduced over time, the residential sector continues to be slightly higher in energy consumption compared to commercial buildings. In the aftermath of the COVID-19 pandemic, the amount of energy consumed in residential buildings is likely higher. Energy consumption variation 100 90 80 70 60 50 (residential vs commercial sector) 40 30 20 10 0 19 19 19 19 19 19 1949 53 57 61 65 69 73 19 19 19 19 19 19 2077 81 85 89 93 97 01 20 20 20 2005 09 13 17 Year Residential Sector Commercial Sector Figure 1-2. Variation in the energy consumption in residential and commercial building sectors (U.S. EIA, [1]) In residential buildings, heating, ventilation, and air conditioning (HVAC) systems consume a significant amount of energy, in order to provide a thermally comfortable indoor environment for occupants. In 2015, air conditioning and space heating systems consumed approximately 17% and 15% of the total electricity, respectively, among residential buildings in the United States [6]. The magnitude of this energy consumption by the HVAC system depends on several factors and their relative importance. Hong et al. [7] suggested that these factors include the occupants and their thermostat preferences, HVAC equipment utilized in a building, and the efficiency of the building and its systems. Energy consuming equipment such as large appliances and electronics comprise another significant portion of energy use. These appliances are also generally used when occupants are present in a building. Final reports from Annex 53 (Total Energy Use in Buildings), which focused on the characteristics that impact overall energy consumption in buildings and associated evaluation methods, include discussions on six key factors which influence building energy 2 consumption: building envelope, appliances, operation and maintenance, indoor comfort, occupant behavior and weather conditions [9]. Among these influential parameters, it was found that occupant behavior, indoor comfort criteria and operation and maintenance of the building and appliances have highest relative impact on the energy use of the building. In another study [10], a study of the correlation between the climate condition, occupant behavior and building types with respect to overall cooling energy consumption found that occupant behavior was the most influential parameter among the three mentioned [10]. In summary, in residential building occupancy and occupant behavior has been found to have a significant impact on energy consumption patterns. The importance of occupant behavior in the residential building sector is more varied, as compared to commercial buildings. In commercial buildings, generally a standard set of institutional energy- consumption rules (i.e., all occupants within the building experience similar indoor environmental conditions resulting from these standard controls) are used to govern how indoor comfort conditions are met [11]. Most commercial buildings operate on regular HVAC operational schedules with a relatively larger number of occupants as compared to residential buildings. These schedules are generally based on the anticipated number of people using the building, and the corresponding ventilation and heating/cooling needs. People in commercial buildings are generally occupying the building for work-related business, and following a fairly regular schedule, particularly when aggregated over many people. By contrast, in residential buildings occupancy and occupant behavior can be much more varied, including the time duration and period occupants spend in their homes and doing various activities [12]. Thermal comfort requirements also vary across by occupant [13]. This leads to a higher uncertainty in energy consumption in residential buildings, as residential buildings are more likely to be adjusted specific to the occupant living in them, rather than set at specific values to please an overall building population. Recent studies also suggest that occupant behavior has a more significant impact on the residential building energy consumption compared to commercial buildings [14-16]. As such, in this research we focus on residential buildings. A significant body of knowledge in recent years has focused on quantifying the uncertainty associated with occupancy behavior in residential buildings, including how occupants’ schedules and interactions with the building impact the overall energy consumption [17]. Several studies 3 found that close to 20% of the overall energy usage can be explained by how occupants interact with the building system where the occupant interaction influences the space heating and space cooling significantly [18,19]. Another study found that the use of plug loads and appliances can be 10 times more variable over time when the appliances’ operation depends on occupant interaction, as compared to when it does not [20]. A different study of a multi-storied housing complex using both cooling and operable windows found that for occupied residential buildings, the air change rate can also vary by up to 87% [21]. Similarly, a study of electricity consumption in 25 residential households in Beijing, China found a variation of 0 to 14kWh/m2 mainly due to occupants’ adjustments of the HVAC system setpoints [22]. A similar study of 305,001 Dutch homes also found significant energy consumption variation [23]. In another study it was shown that the energy consumption can vary by a factor of 3 in similar type of residential buildings only due to occupant behavior [24]. An et al. (2018) studied the energy consumption of air conditioning in different rooms in a residential building in China, obtaining up to 3 to 4 different profiles for different rooms based on the occupant’s utilization of the air conditioning system [25]. In another study, Xia et al. (2019) completed cluster analysis of the energy consumption used by the air conditioning system in residential building in China, finding three different consumption patterns of the air conditioning system, which varied depending on occupant preferences [11]. In summary, all of the above-mentioned studies suggest the importance of occupant interactions with building systems, to support improve prediction and reduced energy consumption in the building sector, and in particular in residential buildings. In addition, both the Annex 66 (Definition and Simulation of Occupant Behavior in Buildings) [26] and Annex 79 (Occupant-Centric Building Design and Operation) [27] are international collaborative efforts focused on the advancement of research and studies of occupant behavior in the buildings. The existence of such efforts further supports the need to study, better understand, and better represent occupants’ impacts on buildings. Occupants, however, are not homogeneous. Their activities and schedules vary substantially, depending on various factors, including demographic characteristics [28]. Unlike commercial buildings, which can be divided into types based on end use, residential buildings all serve a similar purpose. Thus, in current practice, most building energy simulation tools use a predefined fixed occupancy schedule that varies with time of day. This does not vary by or consider the occupants’ 4 potentially diverse characteristics and behaviors and is generally assumed to be the same across the broader U.S. population. For residential buildings, the U.S. Department of Energy (DOE) Reference Buildings [29] and Prototype Buildings [30] utilize a fixed occupancy schedule. This originated from occupancy schedules published in the 1989 version of ASHRAE Standard 90.1 [31] and the Advanced Energy Design Guides (AEDG) studies [32]. Residential building schedules are also provided in the Building America Housing Simulation Protocol [30] and are embedded in the BEopt energy simulation software for residential buildings [33, 34]. These schedules are comprised of several parts, including a numerical representation of the total number of people inhabiting the building, and a 24-hour schedule, which may vary slightly by weekday/weekend and month of the year, which represents the fraction of the total number of people in the building, ranging from 0 to 1. In general, there is little cited information in existing literature about occupancy schedule development used in current Reference buildings and energy modeling software and tools. Some aspects of schedule development are based on energy modeler’s experience and judgement, and generalized assumptions [35]. However, this simplified assumption may lead to significant error in the energy prediction, as discussed by Eguaras- Martinez et al., [36] which indicated that occupant behavior can lead to as high as a 30% difference in the predicted energy use of an energy model. Several recent studies have focused on the development of improved occupancy schedules for buildings. There are several different methods that have been used to develop occupancy schedules, including data-driven methods, machine learning methodologies, and probabilistic methods [37]. Data collection for the schedule development can be done either using real time data collected using sensors, and/or by interviewing occupants. Markov chains are one of the most common methodologies used to develop stochastic occupancy schedules. A simple Markov chain was developed by Page et al. [38] to design the occupancy schedules for a single zone single person building. This model works well when there is a shorter span of absence of occupants in the buildings. This method was then updated for multi-occupant scenarios by Richardson et al. [39]. Another stochastic model was developed by Widen et al. [40] using non-homogeneous time dependent transition probabilities. Building on these models, Lopez- Rodriguez et al. [41] developed a probabilistic model using time use data from Spain from 2009- 2010 [42] to predict the occupancy and energy consumption patterns in residential buildings. 5 Occupancy schedules for two different scenarios, including single zone and multi-zone systems with multiple occupants, were designed using an inhomogeneous Markov chain model [43]. In another study, a stochastic occupancy model was developed using the distribution of time spent and the expectation of a new occupants arriving in the building [44]. The simulation was run for 150 cycles and the generated schedule was used. Using the data collected from a camera and motion sensor, a Markov and a semi-Markov chain model was designed to simulate the schedule for different spaces in a one-day period [45]. In another study, a dynamic Markov chain model was created for 16 commercial buildings where data collection was done for 2 years where several predictors, including season, day of week, time of day and change in window open or close status were used to create the occupancy profiles [46]. In summary, different variation of Markov models have been implemented to evaluate the number of occupants based on the collected occupant information for those spaces. However, these studies generally focus on limited test space types and population types, thus it is challenging to map such results to the overall population of the United States [47]. Data-driven methods have also been used by various researchers to develop occupancy schedules. Cluster analysis is one such method, where occupants are classified into different categories based on their schedule characteristics. Variables such as electricity consumption data have been used as an input to cluster analysis to predict occupancy patterns [48]. Cluster analysis was also used to identify group of similar households based on occupancy profiles for five regions in the UK [49]. In another study, different occupancy patterns were analyzed based on the data obtained from three types of rooms in an occupied building for a time period of five years [50]. Other than cluster analysis, data driven methodologies including Support Vector Machine (SVM) and C4.5 Decision Tree algorithms have been used to predict the occupancy schedules [51]. Artificial Neural Network, Convolutional Neural Network and Recurrent Neural Network are also used to predict representative occupancy schedules [52-54]. A Decision Tree algorithm was implemented to evaluate the presence of occupants in office cubicles using a combination of sensor data like motion, light, acoustic, CO2 and power consumption data [55]. Electricity consumption data of five residential buildings was used to create and compare occupancy schedules using multiple algorithms, Support Vector Machine, K-Nearest Neighbor, and Hidden Markov Model [56]. Decision Tree, Random Forest algorithm was implemented to predict the number of occupants in 6 an open office space [57]. Different methods such as Artificial Neural Network, Support Vector Machine and Hidden Markov Model have been used to predict the number of occupants where CO2 and acoustic data was used as the input to the model [58]. This model has been further extended by adding a Gaussian Mixture Model in the Hidden Markov Model where the objective of the addition of the new model was to categorize changes in sensor data including CO2, acoustic, light and PIR for occupancy detection [59]. Online Extreme Learning Machine was also used in a study to obtain an accuracy of as high as 98% in occupancy detection [60]. Thus, various types of data-driven models have been implemented to predict occupancy information in different studies. The input variables to the model also varies for different test spaces. However, most of these studies focus on assessing occupancy schedules in commercial buildings rather than residential buildings. At the same time, there is no single accepted methodology or defined set of parameters that can be used to predict occupancy scenarios in the residential building sector. In addition, similar to the studies using stochastic methods, the data used to develop the schedules have been collected from specific buildings rather than a broad range of building types. Thus, the resulting schedules vary significantly with the test space considered, and do not represent the overall U.S. population. To collect occupancy information from residential buildings, many sensors are needed [61]. This is because there is not currently a commonly accepted set of variables that can be used to detect occupancy, as discussed in the previous section. The set of variables depend on the test space and the type of activities taking place. However, the use of many sensors can potentially cause significant privacy concerns for some households [11]. To manage this, several studies have focused on collecting the occupancy data using surveys. Balvedi et al. [62] developed a questionnaire-based interview to develop occupancy patterns for both weekdays and weekends. In Italy, a questionnaire-based study was completed across 80 families for two weeks in 2017 to evaluate the percentage of time people spent in their home [63]. A similar study also occurred in China, where the interaction between occupants and energy-consuming appliances was studied using a questionnaire and subsequent interview of occupants [64]. These surveys provide detail insights of occupant behavior and their interaction with the building system. However, to represent the occupancy schedules of the United States population, a dataset needs to be statistically designed represent the overall population, which is expensive and time intensive. 7 Time use survey is a type of dataset which includes people’s activities throughout a 24-hour period, which can be used to estimate the occupancy profiles in different countries. Belgian Time Use survey data [65] was used by Aerts et al. [66] to evaluate a typical occupancy profile in Belgium. Similarly, Blight et al. [67] created weekly occupancy profiles using the United Kingdom time use survey data [68]. Using French time-use survey [69] occupant activities were associated with dummy variables used to evaluate occupant characteristics [70]. A time-inhomogeneous, first order Markov Chain model was used to evaluate the location and activity state of occupant in a residential building using United Kingdom time use survey [71]. For the U.S. specifically, a bootstrap sampling method was used to evaluate occupancy activity patterns in the residential building using American Time Use Survey (ATUS) data [72, 73]. ATUS data from 2009 was used in an unsupervised cluster analysis to evaluate different type of profiles [74]. However, regardless of the methodology used, recent studies neither focused on the development of occupancy profiles for U.S. households with different numbers of occupants, nor have they considered the most recent ATUS datasets and comparisons across multiple years to assess schedule variations over time. Along with the presence and absence of occupants, it is also important to evaluate occupant activity distribution, which is associated with variations in energy used [75]. Several studies have demonstrated the importance of occupant activities on energy-consuming building systems including HVAC, and lighting [76, 77]. Jia et al. [78] showed that occupants’ activities have a large impact on the overall energy performance compared to the presence and absence of occupants. Similar to the activity distribution, it is also important to design the indoor spaces efficiently to meet occupant needs [79]. The importance of studying the occupancy behavior at the temporal and spatial level and the associated activity distributions was discussed by Melfi et al. [80]. In addition, several studies used sensors, including plug load, acoustic and camera-based systems, to study the activity and spatial distribution of occupant in the building systems [81, 82, 83]. However, it is difficult to characterize this distribution as the activity and spatial distribution is multidimensional and depends on various factors [84]. Also, it is important to characterize the activity and spatial distribution of occupants based on their age distribution. In addition, to study these characteristics, significant amounts of data are needed to determine the ground truth scenarios. 8 However, the world population faced a new crisis in recent times due to the presence of global pandemic due to COVID-19 [85]. The pandemic has severely impacted human health and the economic activities throughout the world [86-88]. Several sectors like service and tourism affected substantially due to the pandemic [89,90]. Across the world, a substantial number of people lost their jobs and unemployment rates soared in many countries [91,92]. Preventive measures such as stay-home orders and lockdowns, which were taken to reduce the spread of the virus, also impacted the ways people live [93, 94]. As a significant section of the population began working from home and most K-12 schools and colleges/universities switched to remote learning methods, this resulted in more people staying in home for majority of the day [95-98]. In the U.S., around 35% of the workforce switched to remote working in May 2020, whereas this was approximately 7% before the pandemic [93, 98]. At the same time, as these changes forced people to adapt new living patterns, some of these changes may remain integrated in the people’s lifestyle even after the pandemic has dwindled. Recent surveys showed that, majority of the employees wanted to continue working from home even after the COVID-19 pandemic due to better work-life balance, more flexibility, and less time loss in commute [99-101]. Respondents also preferred to have flexible work schedules, which were more commonly implemented during the work-from-home periods [102-106]. Due to this, as permanent changes in employer expectations of in-person versus remote work schedules have occurred and continue to occur, it will impact the occupancy of residential buildings moving forward, beyond the pandemic. However, there is no study that focused on the impact of pandemic on occupancy schedules and characterize how people from different sector of society adapt their living during this challenging time. 9 2. Research Objectives Given the importance of occupant behavior, particularly in residential buildings, on building energy consumption, as well as the challenges associated with the current practices of building energy modeling, it is important to evaluate the impact of different demographics and socioeconomic characteristics of occupants on their schedules to predict occupancy profiles for residential buildings in the United States. To identify the adapting and changing occupancy scenarios, the impact of pandemic should also be evaluated with respect to different occupant characteristics. The schematic of the focus areas and the flow of this research is shown in Figure 3. Objective 1a: Develop typical Objective 1b: Analyze the Focus area 1: occupancy profiles, and the spatial and distribution of Residential occupancy for impact of demographics on occupants in residential individuals profile variations building Focus area 2: Objective 2: Analyze the occupancy Characteristics of profiles and calculate their occupancy profiles characteristics Focus area 3: Objective 3: Evaluate occupancy Socioeconomic impact on profiles with respect to various Residential occupancy socioeconomic condition Focus area 4: Impact of Objective 4: Analyze the impact of pandemic on Residential pandemic on different section of the occupancy society Figure 1-3. Schematic of the proposed study organization This proposed research consists of four major focus areas. These are discussed in further detail in the following subsections. Focus Area 1: Residential occupancy schedules of individuals The first Focus Area includes the evaluation of occupancy scenarios for individuals. There are several major research questions of focus: a. What demographic factors impact occupancy schedules of individuals? 10 b. How can an occupants’ activities be mapped to the indoor space of a residential building to quantify time spent in different areas of a home? Objective. Develop typical occupancy profiles, and the impact of demographics on profile variations To answer these questions, this section focuses on the development of typical occupancy schedules, and the evaluation of the impact of demographic parameters on these occupancy schedules. There are different parameters that impact occupants’ behavior in residential buildings. The U.S. population is highly diverse, in age, socio-economic status, among others. It is unlikely that a single “typical” occupancy schedule is able to represent the schedules of people of such diversity. This portion of the proposed research thus focuses on the development of typical occupancy schedules, and an assessment of the impact of various demographic characteristics on these occupancy schedules in residential buildings in the United States. This research will also identify typical locations and the types of movements of occupants in residential buildings as a function of time. These are needed to support the design of occupancy sensor systems. Location helps to define where occupants are most, and thus where occupancy sensors systems should focus on for placement. Focus Area 2: Analyze the occupancy profiles and calculate their characteristics This Focus Area aims to evaluate the different types of profiles occupant have in the United States. The main research question of focus is: a. What are the most common types of schedules used by the occupants in the U.S. residential households? How can these schedules be characterized? Objective. Cluster analysis of occupancy schedules in residential buildings in the United States Along with the average profiles, it is also important to identify the mostly utilized occupancy schedules. Thus, the objective of this Focus Area is to analyze the most dominant type of profiles for different demographics of occupants. The percentage of people who followed those schedules are also needed to evaluate to identify the most common type of profiles. The characteristics of these schedules are needed for better insights regarding residential occupancy profiles. This Focus Area is designed to provide detailed characteristics of occupancy schedules to predict the profiles for different type of residential buildings. 11 Focus Area 3: Evaluate occupancy profiles with respect to various socioeconomic condition The goal of this Focus Area is to evaluate the impact of socioeconomic factors of households on their daily schedules. The major research question for this focus area is: a. Does household income impact the residential occupancy profiles, and if yes, how do those profiles vary across income groups? Additionally, how does this occupancy relate to household energy usage? Objective. Variation in Residential Occupancy Profiles in the United States by Household Income Level and Characteristics Occupancy profiles at the individual and household level are studied in Focus Area 1 and 2 respectively. However, along with the demographics of occupants, socioeconomic parameters such as household income also has the potential to impact occupancy profiles. It is beneficial to study the variation in profiles for households at different income levels to support policies and benefits for people who are likely to experience the greatest energy burden. Focus Area 4: Analyze the impact of pandemic on different section of the society This focus area evaluates the impact of the COVID-19 pandemic on different sections of the society. Objective. COVID-19 impacts on residential occupancy schedules and activities in U.S. Homes in 2020 using ATUS The final section of this research focuses on the evaluation of the impact of the pandemic on occupancy schedules and compares them with pre-pandemic. The main research questions answered in this focus area is following: a. Is there any change in total time spent in residential buildings due to pandemic? b. Has the pandemic equally impacted different socio-economic groups? c. How has indoor space utilization changed during the pandemic? Each of these questions is discussed in this section, to support having a better understanding of how occupancy schedules have evolved during the pandemic, and the potential long-term implications of the pandemic on occupancy modeling and residential building energy use. 12 3. Organization This research is organized in eight chapters. Starting with the introduction as Chapter 1, chapters 2, 3, 4, and 5 focus on discussing each Focus Area individually. More specifically, Chapter 2 discusses individual occupancy schedule development, and the impact of varied demographic parameters on occupancy profiles. It also includes an analysis of spatial and activity distribution variation in residential buildings. This chapter is associated with one journal paper (“Typical Occupancy Profiles and Behaviors in Residential Buildings in the United States”, Energy and Buildings, 210, 109713). Chapter 3 discusses the characteristics of profiles for different occupant demographics in the U.S. This chapter is associated with one journal paper (“Cluster analysis of occupancy schedules in residential buildings in the United States”, Energy and Buildings, 236, 110791). In Chapter 4, the impact of socioeconomic parameters on occupancy profiles is discussed and is associated with a journal paper (“Variation in residential occupancy profiles in the United States by household income level and characteristics”, Journal of Building Performance Simulation, 14:6, 692-711). The following chapter is focusing on evaluating the impact of pandemic on the occupancy profiles. This chapter is associated with one journal paper (“COVID- 19 impacts on residential occupancy schedules and activities in US Homes in 2020 using ATUS”, Applied Energy, 119765). The final chapter discusses conclusions, significant research contributions, future studies and followed by the references. The details of publications and references are also given in the final two chapters. 4. Research Organization As mentioned above, this research is organized into four focus areas and each of them resulted in one journal paper for each area. 13 Objective 1a: Develop typical occupancy profiles, and the impact of Journal paper 1: demographics on profile Published in Energy and Buildings Focus Area 1: variations Residential 2020 occupancy for Typical Occupancy Profiles and individuals Objective 1b: Behaviors in Residential Analyze the spatial and activity Buildings 1 in the United States distribution of occupants in residential building Thesis title: Journal paper 2: Evaluation of occupancy Focus Area 2: Objective 2: Published in Energy and Buildings patterns in residential Residential Assess the various type of 2021 building of the United States occupancy for residential households and and the impact of pandemic households evaluate typical household Cluster analysis of occupancy on the profiles (multi-person) distribution schedules in residential buildings in the United States Journal paper 3: Published in Journal of Building Focus Area 3: Objective 3: Performance Simulation 2021 Socioeconomic Evaluate occupancy profiles impact on with respect to various Variation in residential occupancy residential socioeconomic condition profiles in the United States by occupancy household income level and characteristics Journal paper 4: Published in Applied Energy Focus Area 4: Objective 4: 2022 Impact of Analyze the impact of pandemic on pandemic on different section COVID-19 Impact on Residential residential of the society Occupancy Schedules and occupancy Activities in U.S. Homes in 2020 using ATUS Figure 1-4. Diagram of research organization 14 REFERENCES [1] United States Energy Information Administration (US EIA) https://www.eia.gov/totalenergy/data/monthly/ (Accessed October 20th, 2020) [2] EIA, 2020, International Energy Outlook, US Energy Information Administration, https://www.eia.gov/outlooks/aeo/ [3] United Nations Environment Program (UNEP). 2016. Sustainable building and climate initiative https://energies2050.org/sustainable-buildings-and-climate-initiative-unep- sbci/?lang=en (Accessed October 20th, 2020) [4] United States Environmental Protection Agency. 2014. Sources of greenhouse gas emissions. EPA headquarters, Washington, D.C., U.S.A. https://www.epa.gov/ghgemissions/sources- greenhouse-gas-emissions (Accessed October 20th, 2020) [5] Core Writing Team, Pachauri, R. K., and Meyer, L. A. (eds.). 2002. “IPCC, 2014: Climate Change 2014: Synthesis Report. Contribution of Working Groups I, II and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change.” IPCC 40 (1): 151. https://doi.org/10.1046/j.1365-2559.2002.1340a.x. [6] EIA’s Residential Energy Survey Now Includes Estimates for More Than 20 New End Uses, U.S. Energy Information Administration, USA, 2018. [7] Hong T, D'Ocaa S, Turnera, W J. N., C.Taylor-Lange, S, “An ontology to represent energy- related occupant behavior in buildings. Part I: Introduction to the DNAs framework”, Building and Environment, Volume 92, October 2015, Pages 764-777, https://doi.org/10.1016/j.buildenv.2015.02.019 [8] Zhao, H. X., and Magoulès F. 2012. “A Review on the Prediction of Building Energy Consumption.” Renewable and Sustainable, Energy Reviews 16 (6): 3586–92. [9] Yoshino H., Hong T., Nord N., IEA EBC annex 53: total energy use in buildings—analysis and evaluation methods, Energy Build 152 (2017) 124–136, doi: 10. 1016/j.enbuild.2017.07.038 [10] Yun G.Y., Steemers K., Behavioural, physical and socio-economic factors in household cooling energy consumption, Appl. Energy. 88 (2011) 2191–2200, doi: 10.1016/j.apenergy.2011.01.010 [11] Xia D, Lou S, Huang Y, et al. A study on occupant behaviour related to air-conditioning usage in residential buildings[J]. Energy and Buildings, 2019, 203: 109446. [12] K. Amasyali, N.M. El-Gohary , Energy-related values and satisfaction levels of residential and office building occupants, Build. Environ. 95 (2016) 251–263. [13] D. Zhao, A.P. McCoy, J. Du, P. Agee, Lu Y, Interaction effects of building technology and resident behavior on energy consumption in residential buildings, Energy Build.134(2017)223- 233. 15 [14] Santin, O. Guerra, Itard, L., Visscher, H., “The effect of occupancy and building characteristics on energy use for space and water heating in Dutch residential stock”, Energy Build, 41 (2009), pp. 1223-1232, 10.1016/j.enbuild.2009.07.002. [15] Wilke, U., “Probabilistic Bottom-up Modelling of Occupancy and Activities to Predict Electricity Demand in Residential Buildings”. PhD Thesis. 172 (2013), 10.5075/epfl-thesis-5673. [16] Popoola, O. M. 2018. “Computational Intelligence Modelling Based on Variables Interlinked with Behavioral Tendencies for Energy Usage Profile – A Necessity.” Renewable and Sustainable Energy Reviews 82 (May 2017): 60–72. https://doi.org/10.1016/j.rser.2017.09.020. [17] Balvedi, B. F., Ghisi, E, and Lamberts, R. 2018. “A Review of Occupant Behaviour in Residential Buildings.” Energy and Buildings 174: 495–505. [18] Huebner, G.M., Hamilton, I., Chalabi, Z., Shipworth, D., and Oreszczyn, T. 2015. Explaining domestic energy consumption – The comparative contribution of building factors, socio-demographics, behaviors and attitudes. Applied Energy 159(2015): 589-600. [19] Steemers, K., and Yun, G.Y., 2009. Household energy consumption: a study of the role of occupants. Building Research and Information 37(5-6): 625-37. [20] Cetin, K.S., Tabares-Valasco, P.C., and Novoselac, A. 2014. Appliance daily energy use in new residential buildings: Use profiles and variation in time-of-use. Energy and Buildings 84(2014): 716-26. [21] Iwashita, G, and Hiroshi Akasaka. "The effects of human behavior on natural ventilation rate and indoor air environment in summer—a field study in southern Japan." Energy and Buildings 25, no. 3 (1997): 195-205. [22] Li, Z.-J., Jiang, Y., Wei, Q.-P., 2014, “Survey on energy consumption of air conditioning in summer in a residential building in Beijing”, Heat. Vent. Air Cond., 37 (2007), pp. 46-51. [23] D. Brounen, N. Kok, J.M. Quigley, Residential energy use and conservation: eco- nomics and demographics, Eur. Econ. Rev. 56 (2012) 931–945, doi: 10.1016/j. euroecorev.2012.02.007 [24] Fabi, V., Andersen, R.V., Corgnati, S.P., Olesen, B.W., “A methodology for modelling energy-related human behaviour: application to window opening behaviour in residential buildings”, Build Simul J (2013), 10.1007/s12273-013-0119-6 [25] An J, Yan D, Hong T. Clustering and statistical analyses of air-conditioning intensity and use patterns in 457 residential buildings[J]. Energy and Buildings, 2018, 174: 214-227. [26] Yan, D., and Hong, T., 2018. "EBC Annex 66 Final Report-Definition and Simulation of Occupant Behavior in Buildings.". [27] O’Brien, William, Andreas Wagner, Marcel Schweiker, Ardeshir Mahdavi, Julia Day, Mikkel Baun Kjærgaard, Salvatore Carlucci et al. "Introducing IEA EBC Annex 79: Key 16 challenges and opportunities in the field of occupant-centric building design and operation." Building and Environment (2020): 106738. [28] Hong, Tianzhen, Da Yan, Simona D'Oca, and Chien-fei Chen. "Ten questions concerning occupant behavior in buildings: The big picture." Building and Environment 114(2017):518-530. [29] Deru, M., Field K., Studer D., Benne K., Griffith B., Torcellini P., Liu B., et al. 2011. “U.S. Department of Energy Commercial Reference Building Models of the National Building Stock.” Nrel/Tp-5500-46861, no. February 2011: 1–118 [30] Wilson, E., Engebrecht Metzger, C., Hendron, R., and Horowitz S.. 2014. 2014 Building America House Simulation Protocols.National Renewable Energy Laboratory.Vol.NREL/TP-55. [31] ASHRAE. (1989). 90.1 User’s Manual ANSI/ASHRAE/IESNA Standard 90.1-1989. Atlanta, GA: American Society of Heating, Refrigerating and Air-Conditioning Engineers [32] Advanced Energy Design Guides, 2018. [33] Christensen, C., Anderson, R., Horowitz, S., Courtney, A., and Spencer, J. BEopt (TM) Software for Building Energy Optimization: Features and Capabilities. United States: N. p., 2006. Web. doi:10.2172/891598. [34] “BEopt Home,” BEopt, accessed October 10, 2019, http://beopt.nrel.gov/. [35] D’Oca, Simona, and Hong T. 2015. “Occupancy Schedules Learning Process through a Data Mining Framework.” Energy and Buildings 88: 395–408. [36] M. Eguaras-Martínez, M. Vidaurre-Arbizu, C. Martín-Gómez, Simulation and evaluation of building information modeling in a real pilot site, Appl. Energy. 114 (2014) 475–484, doi: 10.1016/j.apenergy.2013.09.047. [37] Saha, H., Florita, A. R., Henze, G. P., Sarkar., S., 2019, “Occupancy sensing in buildings: A review of data analytics approaches”, Energy & Buildings 188–189 (2019) 278–285 [38] Page, J., Robinson, D., Morel, N. and Scartezzini. J. L. 2008. “A Generalised Stochastic Model for the Simulation of Occupant Presence.” Energy and Buildings 40 (2): 83–98. [39] Richardson, I, Thomson M., and Infield D. 2008. “A High-Resolution Domestic Building Occupancy Model for Energy Demand Simulations.” Energy and Buildings 40 (8): 1560–66 https://doi.org/10.1016/j.enbuild.2008.02.006 [40] Widén, J., and Wäckelgård E. 2010. “A High-Resolution Stochastic Model of Domestic Activity Patterns and Electricity Demand.” Applied Energy 87 (6): 1880–92. [41] López-Rodríguez, M.A., Santiago, I., Trillo-Montero, D., Torriti, J., Moreno-Munoz, A., “Analysis and modeling of active occupancy of the residential sector in Spain: an indicator of residential electricity consumption”, Energy Policy, 62 (2013), pp. 742-751, 10.1016/j.enpol.2013.07.095 17 [42] Vale, Raúl Fraguela, Juan José Lorenzo Castiñeiras, and Lara Varela Garrote. "Escuela, familias y ocio en la conciliación de los tiempos cotidianos de la infancia." Revista de Investigación Educativa 29, no. 2 (2011): 429-446. [43] Chen. Z., Xu. J., Soh. Y.C.,,2015, “Modeling regular occupancy in commercial buildings using stochastic models”, Energy Building, 103 (2015), pp. 216-223 [44] Lee C, Tong J, Cheng V., 2014, “Occupant behavior in building design and operation”. In: HK joint symposium 2014, Hong Kong; 2014 [45] Adamopoulou, A. A., Tryferidis, A. M., Tzovaras, D. K., “A context-aware method for building occupancy 477 prediction.” Energy and Buildings 110 (2016) 229–244. [46] Wang, W., Chen, J, Song, X, “Modeling and predicting occupancy profile in office space with a Wi-Fi probe-based Dynamic Markov Time-Window Inference approach”, Building and Environment 124 (2017), 130-142 [47] Schwartz, Z., Uysal, M., Webb, T., Altin, M., “Hotel daily occupancy forecasting with competitive sets: a recursive algorithm”. Int. J. Contemp. Hosp. Manag.,28(2)(2016),pp.267-285. [48] Chen D., Barker, S., Subbaswamy, A., Irwin, D., Shenoy, P, "Non-intrusive occupancy monitoring using smart meters." In Proceedings of the 5th ACM Workshop on Embedded Systems For Energy-Efficient Buildings, pp.1-8. 2013 [49] Buttitta, G., William JN T., and Donal F., "Clustering of household occupancy profiles for archetype building models." Energy Procedia (2016). [50] Vazquez, F. I., and Kastner W. 2011. “Clustering Methods for Occupancy Prediction in Smart Home Control.” Proceedings - ISIE 2011: 2011 IEEE International Symposium on Industrial Electronics, 1321–28. [51] Zhao, J., Bertrand L., Lam, K. P., Yun, R., and Loftness, V., "Occupant behavior and schedule modeling for building energy simulation through office appliance power consumption data mining." Energy and Buildings 82 (2014): 341-355. 490 [52] Lam, K. P., Höynck, M., Dong, B., Andrews, B., Chiou, Y., Zhang, R., Benitez, D., and Choi, J., "Occupancy detection through an extensive environmental sensor network in an open- plan office building." IBPSA Building Simulation 145 (2009): 1452-1459. 493 [53] Amato, G., Carrara, F., Falchi, F., Gennaro, C., Meghini, C., Vairo, C., "Deep learning for decentralized parking lot occupancy detection." Expert Systems with Applications 72 (2017): 327-334. [54] Chen, Z., Zhao, R., Zhu, Q., Masood, M. K., Soh, Y. C., and Mao, K., "Building occupancy estimation with environmental sensors via CDBLSTM." IEEE Transactions on Industrial Electronics 64, no. 12 (2017): 9549-9559. 18 [55] Hailemariam, E., Goldstein, R., Attar, R., and Khan., A., 2011. “Real-time occupancy detection using decision trees with multiple sensor types.” In Proceedings of the 2011 Symposium on Simulation for Architecture and Urban Design (SimAUD '11). Society for Computer Simulation International, San Diego, CA, USA, 141-148 [56] Kleiminger, W., Beckel, C., Staake, T., and Santini. S., 2013. “Occupancy Detection from Electricity Consumption Data. In Proceedings of the 5th ACM Workshop on Embedded Systems For Energy-Efficient Buildings” (BuildSys'13). ACM, New York, NY, USA, Article 10 , 8 pages. DOI=http://dx.doi.org/10.1145/2528282.2528295 [57] Amayri, Manar, Abhay Arora, Stephane Ploix, Sanghamitra Bandhyopadyay, Quoc-Dung Ngo, and Venkata Ramana Badarla. "Estimating occupancy in heterogeneous sensor environment." Energy and Buildings 129 (2016): 46-58. [58] Dong, Bing, Burton Andrews, Khee Poh Lam, Michael Höynck, Rui Zhang, Yun-Shang Chiou, and Diego Benitez. "An information technology enabled sustainability test-bed (ITEST) for occupancy detection through an environmental sensing network." Energy and Buildings 42, no. 7 (2010): 1038-1046. [59] Dong, Bing, and Khee Poh Lam. "Building energy and comfort management through occupant behaviour pattern detection based on a large-scale environmental sensor network." Journal of Building Performance Simulation 4, no. 4 (2011): 359-369. [60] Zou, H, Hao J, Jianfei Y, Lihua X, and Costas S. "Non-intrusive occupancy sensing in commercial buildings." Energy and Buildings154 (2017): 633-643. [61] H. Zou , Y. Zhou , H. Jiang , S.C. Chien , L. Xie , C.J. Spanos , WinLight: a WiFi-based occupancy-driven lighting control system for smart building, Energy Build. 158 (2018) 924–938. [62] Balvedi, B. F., Ghisi, E, and Lamberts, R. 2018. “A Review of Occupant Behaviour in Residential Buildings.” Energy and Buildings 174: 495–505. [63] Carpino, C., Fajilla, G., Gaudio, A., Mora, D., and Simone, M. D. "Application of survey on energy consumption and occupancy in residential buildings. An experience in Southern Italy." Energy Procedia 148 (2018): 1082-1089. [64] Chen, S., Yang, W., Yoshino, H., Levine, M. D., Newhouse, K., and Hinge, A., "Definition of occupant behavior in residential buildings and its application to behavior analysis in case studies." Energy and Buildings104(2015):1-13. [65] Glorieux, I., J. Minnen, T.P. van Tienoven, et al. (2015) Website 'Belgisch tijdsbudgetonderzoek' (www.time use.be), Onderzoeksgroep TOR Vrije Universiteit Brussel & AD Statistiek – Statistics Belgium, Brussel. [66] Aerts, D, Minnen, J, Glorieux, I, Wouters, I, and Descamps, F. 2014. “A Method for the Identification and Modelling of Realistic Domestic Occupancy Sequences for Building Energy Demand Simulations and Peer Comparison.” Building and Environment 75: 67–78. 19 [67] Blight, Thomas S., and Coley, D. A.. 2013. “Sensitivity Analysis of the Effect of Occupant Behaviour on the Energy Consumption of Passive House Dwellings.” Energy and Buildings 66: 183–192. [68] Ipsos-RSL and Office for National Statistics, United Kingdom Time Use Survey [Computer File] (3rd edition), UK Data Archive [distributor] [69] Blanc, M. 2011. INSEE - National Institute of Statistics and Economic Studies - France. http://www.insee.fr/en/default.asp. [70] Wilke U, Haldi F, Robinson D.,2011. “A model of occupants’ activities based on time use survey data” Proceed Build Simul 2011 [71] McKenna, E., Krawczynski, M., Thomson, M., “Four-state domestic building occupancy model for energy demand simulations”, Energy Build., 96 (2015), pp. 30-39, 10.1016/j.enbuild.2015.03.013. [72] Chiou, Y. S., Carley, K. M, Davidson, C. I., and Johnson, M P.. 2011. “A High Spatial Resolution Residential Energy Model Based on American Time Use Survey Data and the Bootstrap Sampling Method.” Energy and Buildings 43 (12): 3528–38. [73] U.S. Bureau of Labor Statistics, American Time Use Survey, http:// www.bls.gov/tus/#data (Accessed on 10th October, 2020.). [74] Diao, L, Sun Y, Chen Z, and Chen J. 2017. “Modeling Energy Consumption in Residential Buildings: A Bottom-up Analysis Based on Occupant Behavior Pattern Clustering and Stochastic Simulation.” Energy and Buildings 147: 47–66. https://doi.org/10.1016/j.enbuild.2017.04.072. [75] Sonta, Andrew J., Perry E. Simmons, and Rishee K. Jain. "Understanding building occupant activities at scale: An integrated knowledge-based and data-driven approach." Advanced Engineering Informatics 37 (2018): 1-13. [76] Galasiu, Anca D., Guy R. Newsham, Cristian Suvagau, and Daniel M. Sander. "Energy saving lighting control systems for open-plan offices: a field study." Leukos 4, no. 1 (2007):7-29. [77] Williams, Alison, Barbara Atkinson, Karina Garbesi, Erik Page, and Francis Rubinstein. "Lighting controls in commercial buildings." Leukos 8, no. 3 (2012): 161-180. [78] Jia, Mengda, Ravi S. Srinivasan, and Adeeba A. Raheem. "From occupancy to occupant behavior: An analytical survey of data acquisition technologies, modeling methodologies and simulation coupling mechanisms for building energy efficiency." Renewable and Sustainable Energy Reviews 68 (2017): 525-540. [79] Roetzel, Astrid. "Occupant behaviour simulation for cellular offices in early design stages— Architectural and modelling considerations." In Building Simulation, vol. 8, no. 2, pp. 211-224. Tsinghua University Press, 2015. 20 [80] Melfi, Ryan, Ben Rosenblum, Bruce Nordman, and Ken Christensen. "Measuring building occupancy using existing network infrastructure." In 2011 International Green Computing Conference and Workshops, pp. 1-8. IEEE, 2011. [81] Zhao, Jie, Bertrand Lasternas, Khee Poh Lam, Ray Yun, and Vivian Loftness. "Occupant behavior and schedule modeling for building energy simulation through office appliance power consumption data mining." Energy and Buildings 82 (2014): 341-355. [82] Kleine-Cosack, Christian, Marius H. Hennecke, Szilárd Vajda, and Gernot A. Fink. "Exploiting acoustic source localization for context classification in smart environments." In International Joint Conference on Ambient Intelligence, pp. 157-166. Springer, Berlin, Heidelberg, 2010. [83] Tomé, Ana, Martijn Kuipers, Tiago Pinheiro, Mário Nunes, and Teresa Heitor. "Space–use analysis through computer vision." Automation in Construction 57 (2015): 80-97. [84] Sonta, Andrew J., Rishee K. Jain, Rimas Gulbinas, José MF Moura, and John E. Taylor. "OESP G: Computational Framework for Multidimensional Analysis of Occupant Energy Use Data in Commercial Buildings." Journal of Computing in Civil Engineering 31, no.4 (2017):04017017. [85] Coronavirus Disease (COVID-19) – Events as they happen., World Health Organization (WHO), https://www.who.int/emergencies/diseases/novel-coronavirus-2019/events-as-they- happen [accessed Sep 29, 2020]. [86]“Coronavirus World Map: Tracking the Global Outbreak”, https://www.nytimes.com/interactive/2021/world/covid-cases.html [87] Recoveries, Managing Divergent. "WORLD ECONOMIC OUTLOOK." (2021). [88] Eduardo Levy Yeyati and Federico Filippini, “Social and economic impact of COVID-19”, June 8, 2021, https://www.brookings.edu/research/social-and-economic-impact-of-covid-19/ [89] Krarti, Moncef, and Mohammad Aldubyan. "Review analysis of COVID-19 impact on electricity demand for residential buildings." Renewable and Sustainable Energy Reviews (2021): 110888. [90] UNWTO. International tourism numbers could fall 60-80%. Madrid, Spain: United Nations World Tourism Organization; 2020. https://www.unwto.org/news/covid-19-international-tourist- numbers-could-fall-60-80-in-2020. [91] Monitor, I. L. O. "COVID-19 and the world of work. Seventh edition, 2020. " Updated estimates and analysis”. [92] Effects of the coronavirus COVID-19 pandemic (CPS), U.S. Bureau of Labor Statistics. Accessed: Sep. 26, 2020. [Online]. Available: https://www.bls.gov/cps/effects-of-the- coronavirus-covid-19-pandemic.htm#. 21 [93] Gibney, Elizabeth. "Whose coronavirus strategy worked best? Scientists hunt most effective policies." Nature 581, no. 7806 (2020): 15-17. [94] Nicola, Maria, Zaid Alsafi, Catrin Sohrabi, Ahmed Kerwan, Ahmed Al-Jabir, Christos Iosifidis, Maliha Agha, and Riaz Agha. "The socio-economic implications of the coronavirus pandemic (COVID-19): A review." International journal of surgery 78 (2020): 185-193. [95] See which states and Cities have Told Residents to stay at home, The New York Times. Accessed: Sep. 19, 2020. [Online]. Available: https://www.nytimes.com/interactive/2020/us/coronavirus-stay-at-home-order.html. [96] J.M. Donohue, E. Miller, COVID-19 and school closures, American Medical Association, JAMA - J. Am. Med. As [97] Moreland, A., Herlihy C., Tynan M.A., Sunshine G., McCord R.F., Hilton C., Poovey J. et al. 2020 "Timing of state and territorial COVID-19 stay-at-home orders and changes in population movement— United States, March 1–May 31, 2020." Morbidity and Mortality Weekly Report 69, no. 35: 1198. [98] Krantz-Kentkrantz, R. M. 2019. "Where did workers perform their jobs in the early 21st century?." Monthly Labor Review: 1-10. [99] Parker, K., Menasce Horowitz, J., Minkin, R., How the Coronavirus Outbreak Has – and Hasn’t – Changed the Way Americans Work, Social & Demographic Trends, Pew Research Center, 2020. Available, https://www.pewresearch.org/social-trends/2020/12/09/how-the- coronavirus-outbreak-has-and-hasnt-changed-the-wayamericans-work/. [100] Barrero, J. M., Bloom, N., Davis, S. J., Why Working from Home Will Stick, Working paper, May 2021; wfhresearch.com [101] Parker, K., Menasce Horowitz, J., Minkin, R., COVID-19 pandemic continues to reshape work in America, Pew Research Center 2022. Available, https://www.pewresearch.org/social- trends/2022/02/16/covid-19-pandemic-continues-to-reshape-work-in-america/ [102] Louie, D., “How COVID-19 pandemic forever changed the workplace, what companies are doing going forward”, ABC7NEWS, March 17, 2021 [103] Weekes, W., “13 ways the coronavirus pandemic could forever change the way we work”, CNBC, April 29, 2020. [104] Mudditt, J., “How offices will change after coronavirus”, BBC, May 14, 2020 [105] Davenport, T. H., Redman, T. C., “Experiments and data for 796 post-COVID-19 work arrangements”, MITSloan Management Review, March 23, 2021 [106] “How the COVID-19 pandemic is changing the way we work”, Corehealth Technologies, Jul 16, 799 2020 22 2. CHAPTER 2 - TYPICAL OCCUPANCY PROFILES AND BEHAVIORS IN RESIDENTIAL BUILDINGS IN THE UNITED STATES Mitra, Debrudra, Nicholas Steinmetz, Yiyi Chu, and Kristen S. Cetin. "Typical occupancy profiles and behaviors in residential buildings in the United States." Energy and Buildings 210 (2020): 109713. 1. Abstract The energy performance of a residential building is highly dependent on occupant’s presence or non-presence in a building and their interactions with energy-consuming appliances. Typical occupancy schedules for residential buildings must be defined for applications such as building energy modeling as well as for assessing energy savings associated with the use of occupancy sensing technologies and occupancy-dependent controls. Currently, commonly used simulation programs assume a typical occupancy schedule, however, there is significant opportunity for improvement to these schedules as this is generally based on engineering judgement. This research uses 12 years of the American Time Use Survey (ATUS) data to develop typical occupancy schedules for a range of household types and occupant age ranges. This is compared to currently utilized residential occupancy schedules. In many cases the developed schedules exhibit similar patterns, however, differences are also found to be as high as 41% for certain periods of time. For occupancy sensing applications, the spatial-temporal distribution of occupants in residential buildings is also evaluated. These locations vary based on temporal factors as well as demographic factors such as age and number of occupants. The results of this research work towards improved occupancy schedule development can benefit both industry professionals and researchers. 2. Keywords Occupancy schedule, Presence, Residential building, Spatial distribution, American Time Use Survey 23 3. Introduction The building sector is one of the largest energy consuming sectors throughout the world, 41% of which originates from buildings. This energy use is expected to continue to increase moving forward (Cantin et al. 2012, US EIA 2010). In residential buildings in the U.S., which represent approximately half of the U.S. building energy usage, the heating, ventilation and air conditioning (HVAC) system designed to meet the comfort requirements of occupants, consumes approximately 51% of annual energy use on average (US EIA 2015). The magnitude of this HVAC consumption is dependent on the equipment utilized in homes, and on occupants and their thermostat preferences, as discussed by Hong et al. (2015). In addition, the use of most of the remaining energy consuming appliances are also highly dependent on the occupants’ level of use and their behaviors (Zhao et al., 2012). As such, occupants in residential buildings represent a significant source of uncertainty in energy consumption (Popoola, 2018). For example, as was shown by Iwashita and Akasaka (1997), due to variations in occupants’ behavior, ventilation rates in homes can vary by up to 87%. Another study found that in summer, the residential air conditioning electricity consumption across 25 households varied from 0 to 14 kWh/m2 (Li et al., 2014). Fabi et al. (2013) showed that the energy consumption can vary by a factor of 3, only due to occupant behavior in similar types of residential buildings. As such, that there is significant uncertainty in the energy performance of the buildings due in large part to occupant behavior. In International Energy Agency (IEA) Annex 53, occupant behavior has been selected as an important parameter for evaluating energy performance (Yoshino et al., 2017); in addition, Annex 66 (Yan et al., 2018) and 79 both focus on occupant behavior in buildings. Thus, it is important to work towards improvements in the understanding, characterization, and modeling of occupancy in buildings. Most residential energy modeling studies use existing occupancy schedules commonly used in publicly available energy modeling software programs. For residential buildings, the U.S. Department of Energy Reference Buildings (Deru et al. 2011) and Prototype Buildings (Wilson et al. 2014) utilize an occupancy schedule which originated in part on occupancy schedules based on schedules published in the 1989 version of ASHRAE Standard 90.1 (ASHRAE 1989). Residential building occupancy schedules are also provided in the Building America Housing Simulation Protocol (Wilson et al. 2014) and are embedded in the BEopt energy simulation software for 24 residential buildings (Christensen et al., 2006, BEopt, 2019). These schedules generally consist of several components, including the maximum number of people which could inhabit the building, a 24-hour hourly base schedule ranging in value from 0 to 1 (0 is no occupants and 1 is maximum occupancy), and multipliers which may be used to vary the base schedule slightly by factors such as weekday/weekend and month of the year. In general, there is little cited information in existing literature about the development of the occupancy schedules used in these reference buildings and energy modeling software and tools. Some aspects of schedule development are based on energy modelers’ experience and judgement, and generalized assumptions (D’Oca et al. 2015). As such, this work focuses on occupancy schedule development from other data sources, and a comparison to existing assumptions in these software tools. Such schedules are important particularly for assessing the energy savings associated with the implementation of HVAC-connected occupancy sensor systems. Several recent studies have focused on improvements in occupancy predictions and studying how occupant behavior impacts energy consumption. A broad overview of different methods used in occupant behavior studies is included in Saha et al. (2019). Recent studies can generally be divided into those that use existing data and/or datasets to develop scheduling using probabilistic, data- driven, and/or machine learning methods, and those studies which focus on collecting new occupancy data, either through sensor data collection or through interviews to collect qualitative and quantitative data. For those that focus on occupancy data analytics and model development using existing data, time use survey data from the country of study is commonly used. For probabilistic models, Markov chains are one of the more common methods used to stochastically model occupancy and predict occupancy profiles. One of the initial efforts using this approach was by Page et al. (2008), where a simple Markov chain was applied to design a single zone system for a single occupant. This approach was then extended for use with multiple occupants in Richardson et al. (2008). Widen (2010) also developed a stochastic model based on time series data, where the transition probabilities were calculated based on non-homogeneous time dependent transition probabilities. Combining these methods, Lopez-Rodriguez et al. (2013) developed a probabilistic model to evaluate the occupancy and energy consumption pattern in the residential building sector based on the Spanish time series data of 2009-2010 (Vale et al.,2011). Aerts et al. (2014) developed an 25 occupancy probabilities model using the Belgian Time-Use survey data (Glorieux et al., 2015). Blight et al. (2013) created weekly occupancy profiles using the UK time use survey data. Vazquez et al. (2011) used a clustering algorithm to identify occupancy patterns based on the data of three types of rooms collected from an occupied building over five years. Chiou et al. (2011) captured the occupant activity patterns in the ATUS data using the bootstrap sampling method. Stochastic models were developed by Wilke at al. (2011) using the French time-use survey (Blanc et al., 2011) where the occupant activities are associated with dummy variables to evaluate different characteristics of the occupants. Mckenna et al. (2015) used the UK time use survey data (Ipsos- RSL) and created a time-inhomogeneous, first order Markov Chain method to evaluate the location and activity state of occupant in a residential building. The American time use survey (ATUS) (U.S. BLS 2009) has been used to model residential occupant behavior using unsupervised cluster analysis (Diao et al. 2017). Chen et al. (2019) studied the different energy consumption patterns for occupant groups of different income ranges in residential buildings using the 2016 American time use survey data. However, regardless of the methodology used, recent studies have not focused on the development of occupancy profiles for households with different numbers of occupants, nor has any study looked at the most recent ATUS datasets and comparisons across multiple years to assess schedule variations over time. Apart from creating a stochastic model, several studies have focused on collecting new building occupancy data using interview methods or using data collected from sensors installed within a building. For example, a recent study by Balvedi et al. (2018) developed questionnaires and conducted interviews used to evaluate the occupancy patterns to develop a 24 hours occupancy profile for weekdays and a 48 hours profile for weekends. Wang et al. (2017) also studied occupancy in commercial spaces for 5 weekdays, 1 holiday and 1 weekend for occupants who carried a WIFI emitting tag. Wagner et al. (2018) listed different technologies used to monitor occupancy behavior, including the application of image based, motion based, indoor environmental parameter based, and WIFI based sensors. In Hailemariam et al. (2011), the combination of PIR, light, acoustic, CO2 and power consumption data were incorporated into a decision tree algorithm to evaluate the presence of occupancy in an office cubicle. A detection accuracy of 98.4% was reported when acoustic data was also used as the input in the analysis. Kleiminger et al. (2013) used electricity consumption data of 5 buildings for 8 months to detect 26 occupancy in residential building, then assessed the performance of learning algorithms, including support vector machine (SVM), k-nearest neighbor (KNN) and Hidden Markov Model (HMM). In another study, occupancy detection in an apartment was evaluated based on the trajectory of indoor climatic data, including PIR, acoustic, air temperature, CO2 and VOC values (Pedersen et al., 2017). A high accuracy of more than 98% was reported based on using Online Extreme Learning Machine (OS-ELM) (Zou et al., 2017). These studies however, are generally location- specific, in that the data is collected from one or several particular buildings, rather than a broad range of buildings. As such, it is likely that changes to the test space would impact the accuracy of occupancy detection. In summary, several probabilistic and data-driven methodologies have been completed in recent years to evaluate the occupancy levels in buildings. However, there are limitations of the current methods in occupancy prediction. First, most studies focus on the commercial buildings, whereas there are comparatively fewer studies on residential buildings. For those studies that have focused on residential buildings, none have considered the differentiation of schedules for different numbers of household members or the household member makeup. In addition, those that have developed schedules using data-driven methods generally have used building-specific data rather than data to characterize the overall building stock in the U.S. As such, it is challenging to generalize these conclusions to “typical” households. Therefore, the main objective of this study is to develop and characterize typically occupancy schedules for residential buildings in the U.S. based on household size, as well as other influential factors. In addition, this effort then extends to analyze the spatial distribution of occupants in a residential building, as it relates to the developed schedules. To do this the American Time Use Survey (ATUS) data was used as the main data source. The results of this work help to better understand occupant use of residential buildings, including identification of influential factors impacting residential occupancy schedules, for applications in energy modeling, as well as occupancy sensor system performance evaluation. 4. Occupancy Data Two main datasets were used in this work including the American Time Use Survey (ATUS) data and the Residential Energy Consumption Survey (RECS) data. These two datasets are developed to statistically represent the overall U.S. population. Each are summarized as follows. 27 4.1. American Time Use Survey (ATUS) The ATUS dataset is a survey supported by the U.S. Bureau of Labor Statistics (U.S. BLS), which is conducted annually by the United States Census Bureau. The survey compiles national-level measurements of the amount of time that people living in the U.S. spent doing various activities. From 2003 to 2018, this includes data from over 190,000 interviews. Data has been collected using the same methodology throughout this time period, enabling the data collected across many years to be utilized together and compared over time. This data is collected via in-person, telephone, or email interview where one member over 15 years of age from a pre-selected household is asked to discuss the activities, they completed over a span of 24 hours, starting from 4:00 am of the previous day. The survey collects data on the activities conducted by the person, their duration in as small as 5-minute increments, the occurrence of the activity on a weekday or weekend, and the presence of other people during the activity. A pre-defined list of over 18 major activities and 461 detailed activities is used to match the described activity to the appropriate code. Activities include a broad range of category such as sleeping, housework, care for children, work, education, and entertainment. For each individual interviewed, details about their age and gender are also collected within the ATUS data. The selection of the households is completed based on those households that recently completed the Current Population Survey (CPS), another survey which is also governed by the U.S. BLS. The CPS includes demographic data which can be linked to the ATUS data for further information on the households surveyed. A weighing function is used for each person interviewed to enable a statistical representation of the U.S. population. For this study, the ATUS data of each year, starting from 2006, the occupant ID, age, activity, location and the weightage factors value were collected and combined across the 12 years to make the final comprehensive dataset. The age of people in different households have also been collected for the specified time span. 4.2. Residential Energy Consumption Survey (RECS) The RECS survey focuses on collecting data about the characteristics and whole-home and energy use energy consumption of residential buildings throughout the U.S. Data is collected to support an understanding of these characteristics as the climate zone and geographic regional level (U.S. EIA 2015). This study, administered by the U.S. Energy Information Administration, began in 1978; since this time data has generally been collected every 6 years via survey. In this research, 28 the data on the age distribution of households of different sizes is used from the RECS data in 2009 and 2015. 5. Methodology This study is divided into three parts in order to develop occupancy profiles of residential buildings based on ATUS and RECS data, including the development of overall occupancy schedules, household occupancy schedules, and in-home spatial distributions of occupants. In summary, in this work the overall average occupancy schedule of people’s presence or non-presence in a home is determined, which then is extended to evaluate the average occupancy schedule for households with different numbers of occupants. To evaluate the typical occupant characteristics in households, the correlation of occupant ages in different residential buildings is next evaluated. Finally, the spatial distribution of occupants in an indoor residential building space is studied and compared. In the ATUS data, the location and activities of people is given for a 24-hour period. However, for some of the activities, the location is not specified. In this situation, a location is assigned based on the activity description. Activities are first classified as those that are within a home or outside of a home. For those within a home, a specific space within the home is defined (e.g. living room, kitchen, bedroom) based on the likely location of occurrence of that activity. For example, for sleeping, grooming (e.g. brushing teeth or hair) and personal activities, the location of these activities is typically not mentioned in the activity description. As such, it is assumed that if sleeping is occurring, it is likely in a home, in the bedroom. Similarly, for grooming and personal activities, if the person is already at home and these activities are conducted, they are still assumed to be in the home. An example of this mapping is included in Table 1. 29 Table 2-1. Examples of activity data mapping to overall and spatial location in residential buildings Example activities given in ATUS Presence in residential Spatial location in building building Work, main job No (given) -- (absent from home) Eating and drinking Yes (given) Dining room Television and movies Yes (given) Living room Washing, dressing and grooming Yes (assumed) Bathroom oneself Socializing and communicating with Yes (given) Living room others Sleeping Yes (assumed) Bedroom After the establishment of this linked data, the schedule for each person in the ATUS dataset over the 24-hour period is mapped to reflect a binary value (0 = not present in home, 1= present in home) over 5-minute time intervals, where it is assumed that during that interval, the person completes the same activity across the entire 5-minute period. Given the 5-minute granularity of the ATUS data, a higher level of granularity was not feasible. In the ATUS dataset, among the occupant related variables, including age, gender, financial information, etc., which can potentially impact the occupancy schedule in a residential building, age is considered in this study. The reason for selecting age as an important variable for occupancy study is that it can be mapped to occupant characteristics of different types of households. Based on the ATUS and RECS data, the typical age combinations of different types of households are estimated, which are used evaluate the occupancy schedule for different types of households. The age of the occupants is divided into seven age groups, including under 25, 25 to 34, 35 to 44, 45 to 54, 55 to 64. 65 to 74, and over 75 years. These divisions are chosen to be consistent with the RECS data age divisions. Weekdays and weekend designations are also used to divide the data, consistent with previous literature that indicated significant differences in occupancy schedules on weekdays/weekends (Balvedi et al., 2018). For each of the variables, the ATUS dataset is studied individually, and an average typical schedule is created. An overview of the data analysis methodology for the development of schedules by occupant age and weekday/weekend designation is included in Figure 1, where F1 and F2 represents the result of the occupancy schedule for individual occupants (F1) and a household (F2) respectively. 30 Figure 2-1 Overall framework of occupancy scheduling in residential building Using this method, the schedule for each combination of variables is developed using both the ATUS and RECS datasets. According to RECS data, there are five different household size designations, including, 1-, 2-, 3-, 4- and 5+ members. Typical schedules have been created for all the different types of buildings which represents the majority of the residential building in the US. The variation in time of absence for occupants with different age group have also been studied. After the overall schedules for the typical residential buildings is evaluated, the spatial distribution of occupants in different building sectors are evaluated. 6. Results and Discussion This section includes the results and discussion associated with (a) typical occupancy schedules in different age groups, (b) the number of hours of absence in the home, (c) occupancy schedules for typical U.S. residential household types, and (d) the spatial location distribution of occupants in residential buildings. The combination of the results of the four sections represents the overall occupancy profiles in residential buildings in the United States. 6.1. Typical occupancy schedule in different age groups Fourteen residential occupancy profiles were created representing age-based schedules on both weekends and weekdays. Prior to creating the average schedules across all twelve years of data, each individual year was also evaluated to ensure that overall trends and time interval were consistent from year to year; no outliers were found to exist. In order to compare year to year 31 occupancy schedules, the standard error between the occupancy profiles was calculated and found to be not statistically significant for each age group on both weekdays and weekends, as shown in Table 2. The relative variation of the average of the data for each of the 12 years can be seen from this calculation, where the small value of the standard error in Table 2 indicates that the profiles obtained from the ATUS data is consistent across the years studied. Table 2-2. Standard error of average occupancy schedules across 2006-2017 ATUS data, by age and weekday/weekend Under 25 25-34 35-44 45-54 55-64 65-74 Over 75 Weekdays 0.0005 0.0002 0.0002 0.0003 0.0002 0.0003 0.0003 Weekends 0.0008 0.0002 0.0002 0.0002 0.0001 0.0002 0.0002 The average of all twelve years of ATUS data is then used to create unique occupancy schedules. Figure 2 displays the occupancy presence profiles in a residential home of each age group on both weekdays and weekends. A ‘1.0’ for occupancy fraction indicates that all occupants surveyed in this group reported being in a home, while a ‘0.0’ indicates that no occupants reported being in a home. It can be seen from Figure 2 that significant differences exist between weekday and weekend schedules and that schedules also vary by occupant age. There are a number of observable trends, some of which are not represented in the current standard occupancy schedules for residential buildings. On both weekdays and weekends in the early mornings, nearly all occupants remain in their homes until roughly 6:00am. As individuals began to leave the household, differences can be seen in occupancy fractions among the different age groups. For younger and working-age groups in the United States (i.e. less than 65), the rate and percentage of occupants that leave their homes (56-68% in weekdays and 38-44% in weekends) is much greater than that for older age groups (30-38% in weekdays and 28-36% in weekends). Additionally, the rate and percentage of occupants that leave their homes in the morning is significantly less on weekends, particularly for younger age groups. 32 Figure 2-2. Average occupancy (%) in residential buildings by age group for (a) weekdays and (b) weekends in comparison to the schedules used in the residential (multi-family) DOE Reference Building and Building America Simulation Protocol Figure 2 shows that on weekdays, for people in age groups 55-64 and under, the occupancy fraction reduces significantly until approximately 9:00 am and then remains nearly constant until approximately 3:00 pm. After this time, people begin to return to their homes and the occupancy fraction increases. The rate of increase of the occupancy fraction is much lower compared to the reduction rate in the morning. This indicates that people normally leave their home at similar times, but the returning time varies more significantly by person. On weekends, the profile of the occupancy fraction is more similar for people across all age groups compared to weekdays, though the minimum value of the occupancy fraction is lower for those under the age of 55. People in the age groups 25 to 34, 35 to 44, and 45 to 54 also display similar profiles on weekends with the difference between the maximum and minimum occupancy fractions are being 0.396, 0.387, 0.411 and 0.390 respectively. The occupancy fraction profile for those over 65 is nearly the same as weekdays. The occupancy fraction for those under 25 is higher throughout the day before 5:00 pm compared to those 25 to 74. For the remaining part of the day, it is more common for those in the middle age group to remain in their home compared to those under 25. People under 25 tend to be likely to leave their home earlier in the day compared to those in other age groups. In comparing weekdays and weekends, there are several notable differences. The minimum occupancy fraction for all the age groups reaches the minimum of around 0.3 on weekdays whereas it is approximate 0.7 in weekends. Among all age groups, those 25 to 34, 35 to 44 and 45 to 54 33 have the lowest occupancy fractions throughout the middle part of the day, which indicates that a comparatively larger amount of people in these age groups are not present in their home during the daytime. During weekends, the occupancy variation among different age groups is lower, with an occupancy fraction difference of approximately 0.2, whereas the difference is much higher during weekdays, with an occupancy fraction difference of approximately 0.4. In addition, the occupancy fraction remains at a minimum level for most of the age groups from approximately 10 am to 3 pm, whereas, on weekends, the occupancy fraction value reaches its minimum only for a very small timespan at approximately 12 pm. The average occupancy profiles created herein are also compared with the profile utilized in the DOE Reference Building (multi-family residential) and Building America (BA) protocol. From Figure 2, we can conclude that the occupancy profiles for different age groups vary significantly by age group as well as weekdays and weekends. However, both the DOE Reference Building schedule and BA protocol use a single averaged profile over all the age groups for both the weekdays and weekends. The occupancy fraction in theses reference buildings and protocols is similar to the profiles obtained from this study, however, the occupancy fraction value is overestimated in the morning from around 5:00 am to 8:00 am and underestimated from approximately 7:00 pm to 10:00 pm. Similarly, the reference profile underestimates the occupancy fraction during the daytime. The profiles obtained from the ATUS study in this paper could be used to update the average profiles utilized in the DOE Reference Building and BA Protocol. In addition, which could benefit from improved accuracy in energy modeling, and in evaluating energy saving potential of energy-consuming devices, particularly those that are occupancy-based. 6.2. Number of hours of absence from the house ATUS data was next used to evaluate the typical durations of absences for four age groups, including under 25, 25-54, 55-65, and over 65, on both weekdays (Figure 3) and weekends (Figure 4). These results are critical for the development of occupancy schedules for use in evaluating the impact of occupancy-based HVAC controls on energy savings. To evaluate this, the length of time that occupants are absent from a home is needed, as these absence distributions impact an understanding of the presence versus absence profiles of occupants, to support energy savings evaluation. We also note that, in a residential setting, the complete absence of all occupants is more 34 important than in larger commercial buildings, as the capabilities of HVAC systems in homes in the U.S. are limited, typically to an on or off state. Therefore, complete absence supports an occupancy-based control strategy where the HVAC system operates at a setback condition and achieves energy savings as compared to a constant setpoint temperature regardless of occupancy. Figure 2-3. Hourly distribution of the absence profile in weekdays of occupants age group (a) under 25, (b) within 25-54, (c) within 55-64 and (d) over 65 years It can be seen that on weekdays, there are significant age-based differences in how long occupants are absent from their home. In general, the most common schedule for every age group is those who stay at home for the entire day, i.e., zero hours of absence, ranging from 11% to 32% of people in each age group. For those under 25, and between 25 and 55, the most common time away from home ranges from 8 to 12 hours, representing approximately 42% to 44% of the occupants’ schedules across these durations. Likely, these absence periods align with work-driven schedules and school-driven schedules that are common for individuals of these ages. As the age range increases, occupants begin to spend more time at home. For people ages 55 to 65, there is still a visible peak in the distribution centered around 10 hours of absence per day. In comparison to the younger age groups, however, only 33% of occupants’ schedules are represented by absences of 8 35 to 12 hours. This is likely because individuals typically retire from working, and thus at least due to work-related commitments, spend less time away from home. For occupants over 65 years of age, over 30% report remaining at home throughout the day, which in comparison to the other age groups (12%, 12%, and 20%) is significantly higher. In addition, unlike the other age groups, there is no peak in hours away from the home in the 8- to 12-hour range; instead, as the number of hours absent per day increases, the percentage of people following such a schedule decrease. A similar analysis was done for weekend schedules. Figure 4 shows the absence distributions for the same four age groups on weekends. Figure 2-4. Hourly distribution of the absence profile in weekends of occupants age group (a) under 25, (b) within 25-54, (c) within 55-64 and (d) over 65 years While the differences in absence distributions among age groups is less prevalent on weekends, there is a clear distinction between distributions on weekdays as compared to weekends. On weekdays, there is a peak in length of absence of 8 to 12 hours per day for those in age groups below 25, and 25 to 54; on weekends, however, it is much less common for occupants to be absent from their homes for large portions of the day. Occupants normally do not follow the same work or school schedule on the weekends, which is supported in these results. Additionally, it can be 36 seen that for all age groups, it is more common (over 20%) for people to remain in their home throughout the entire day. All age ranges show a similar trend in the percentage of people that the most common profile is being present in their home for majority time periods of day. Overall, the results show the importance of considering age and whether it is a weekday or weekend when quantifying occupant behavior. 6.3. Occupancy characteristics for typical residential buildings in the U.S. Currently, energy simulation tools use the same profile for all types of households irrespective of the number of members in every home. In this study, the homes were divided based on the number of household members to study the differences of the occupancy profiles among each household size. Utilizing the data obtained in the previous section, an overall occupancy profile is next developed for four categories of households, following the divisions designated in the RECS datasets. These four categories include households with: (a) one, (b) two, (c) three and (d) four members. The five or more member homes are not included in this study as the total number of occupants in this category is not known, and this constitutes less than 10% of the people in the overall U.S. population, a smaller fraction as compared to the other household types. Among the four types of homes, homes with two members are the most common. Based on the distribution of resident age groups in each of the types of homes (Table 2), a typical combination of age group distributions for different types of homes is determined. For the 2006 to 2008 ATUS data, the RECS 2009 (US EIA, 2009) survey was used as this data contains the occupant age distribution data from 2003 to 2008. For the remaining years, the RECS 2015 data (US EIA, 2015) was used. For a single-member household, the occupant age is most commonly over 75, whereas for two-member households, the most frequent ages are both in the 55 to 64 age range. For three- person households, the three most common age ranges are two adults ages 25 to 44 (age groups 25 to 34 and 35 to 44), and one child or young adult under 25. For the four-person household, the most common are two working-age adults ages 25 to 44 and two children or young adults under 25. 37 Table 2-3. Percentage of occupants by age group and household size from the RECS 2009 and 2015 (US EIA) datasets 1-member 2-members 3-members 4-members 5 or more members home home home home home Age Group 2009 2015 2009 2015 2009 2015 2009 2015 2009 2015 Under 25 3.19 3.48 5.03 4.22 7.18 8.76 4.46 5.16 3.94 5.93 years 25 to 34 10.22 9.76 12.29 13.11 20.44 19.07 22.93 20.65 23.62 22.88 years 35 to 44 10.54 8.36 10.06 7.03 21.55 19.59 35.67 32.26 40.16 37.26 years 45 to 54 17.57 13.24 19.83 15.22 27.07 24.74 25.48 25.81 22.05 21.19 years 55 to 64 21.09 25.09 25.48 26.00 14.92 16.49 8.28 10.32 7.87 10.17 years 65 to 74 15.34 19.86 16.76 21.55 5.52 9.28 2.55 3.23 1.57 1.69 years Over 75 22.04 20.21 11.45 12.88 3.31 2.06 0.60 1.94 0.80 0.80 years To further verify the age ranges of the occupants in multi-person households, the ATUS data is used, which provides the ages of occupants in a household. As the RECS data does not provide information on correlations of ages of household members. A correlation matrix has been evaluated between all combinations of ages of occupants for 2-, 3- and 4-member households. The correlation matrix for 2-member households is shown in Table 4, where the cells in this table represent the percentage of 2-member households within those age groups. The coefficients have been calculated using the following equation 𝑋!" 𝑥!" = ∑!$% ∑#"$% 𝑋!" # Where, i, j are the age groups, 𝑋!" is the number of people belonging to ‘i’-th age group, where the other person in the household is of age group ‘j’. The higher the number of each of the elements in Table 4, the higher the probability the occupants belong to that particular age group. As it is shown in the results, the maximum value occurs for a 2-person household where the age of both the occupants are 55 to 64. This is consistent with the RECS data, where the most common ages of 2-member households are where both of the occupants are 55 to 64. 38 Table 2-4. Correlation matrix of occupant ages for 2-member households, based on ATUS data Age Under 25 25-34 35-44 45-54 55-64 65-74 Over 75 Under 25 2.20 4.24 4.72 5.41 2.20 0.44 0.21 25-34 5.93 1.86 0.84 1.11 0.48 0.14 35-44 2.44 2.31 1.03 0.65 0.29 45-54 6.70 6.43 1.52 1.21 55-64 14.43 7.88 1.87 65-74 11.56 5.11 Over 75 6.76 A similar analysis is performed to evaluate the correlation matrix for 3- and 4-member households. For the 3 and 4- member households, all age combinations of household members have been evaluated, which resulted in 84 and 210 matrix components respectively. Among those elements, the top three combinations are shown in Table 5. Table 2-5. Most common age combinations for 3- and 4-member households, based on ATUS data 3-member household Major age combinations Percent of households (%) 1 people under 25, 2 people 25-34 12.2 1 people under 25, 2 people 45-54 12.1 1 people under 25, 2 people 35-44 11.0 4-member household 2 people under 25, 1 25-34 & 1 people 35-44 25.7 2 people under 25, 2 people 35-44 25.7 2 people under 25, 2 people 45-54 15.2 Based on the distribution of ages in each size household, the most common combination of age group distributions is used in this research for the creation of typical occupancy schedules. These are as follows. For a single-member household, over 75; for a 2-member household, both occupants are 55 to 64; for a 3-person household, two people 25 to 34, and one under 25; for a 4- person household, based on the RECS and ATUS data, one-person age 25 to 34, one 35 to 44, and two others under 25. 6.4. Occupancy schedules for typical residential buildings in U.S. A schedule for the four different types of households for both weekdays and weekends are next created by combining the average schedules of each of the occupant types together (Figure 5). This is used for comparison with existing occupancy schedules used in the DOE Reference Building 39 (Deru et al. 2011). The profile used in BA Simulation Protocol (Wilson et al. 2014) is also very similar to the schedule used in the Reference Building. All occupancy schedules follow similar trends; however, there are also noted differences between the profiles. The 1-person household has the highest occupancy fractions throughout the day on both weekdays and weekends; this is likely due to the age of the occupant being older and thus less likely working outside of the home. The occupancy for the 2- and 3-member households are overall lower during the day as compared to the 1-member household; the two household types are also highly similar on both weekends and weekdays. Slight differences occur in the morning around 8:00 am, where the 3-member household occupancy decreases more quickly during weekdays. For both weekday and weekends, the occupancy fraction for the 2-member household is higher for the first half of the day compared to the 3- and 4- member households. In comparison to the occupancy schedules used in the existing DOE reference buildings (Deru et al. 2011) and in BEopt (2019), these ATUS-based schedules show some significant differences. The existing occupancy profile overestimates the occupancy in the morning and the latter half of the day and underestimates the profile in the daytime for all household types, in comparison to the results found in this analysis. For parts of the day, the reference building occupancy schedule is similar to several of the typical household occupancy schedules, whereas the deviation is as high as approximately 45% on weekdays and 44% on weekends. 40 Figure 2-5. Occupancy (%) for different types of households, in comparison to residential DOE Reference Building on both weekdays and weekends 6.5. Spatial location distribution of occupant in the building system Next, based on the locations assigned or designated for each of the ATUS data-specified activities (Table 1), the spatial distribution of the occupancy profiles is also assessed for each age group on weekdays and weekends, as shown in Figure 6. The locations within the home considered include the following: (1) bedroom, (2) bathroom, (3) living room, (4) dining/kitchen, (5) office, (6) other (laundry, gym, etc.), and (7) garage. For all ages of occupants, irrespective of day of the week, they spend majority of the time in the bedroom when at home. During the day, when people are at home, it is most common to spend this time in the living room. The time spent in the living room also increases with the age range of the occupant. On weekdays, people normally stay in their living room in the evenings, whereas, on weekends, time spent in living room is almost uniform in both the morning and evening. In terms of kitchen and dining use, except for those older than 65, on weekdays the dining/kitchen area profile is more used in the early evening, around the time when dinner would be made and eaten. However, on weekends, there are two common times of use, including one mid-day during lunch, and another early evening for dinner. This is likely 41 because households appear to normally have both their lunch and dinner in their home on weekends. It is also noted that people under 25 years of age, on average, spend more time in an office area compared to other age groups, likely due to either school related-work or studying. For people over 65, the weekday profile is similar to the weekend profile. Several peaks can be seen in the profile for bathroom spaces in the morning but diminishes with the daytime. Time spent in the other and garage spaces remains relatively small, and nearly almost uniform for all age groups on both weekdays and weekends. Figure 7 shows the percentage of time, of the total time spent in the home, that occupants in each age group spent in the seven different interior space types. These results show that occupants spend most of the time in their home, in the bedroom, as previousy mentioned, mostly sleeping. For the remainder of the time, the living room area accounts for nearly half of the occupants’ time spent outside of the bedroom. The percentage of time spent in the bedroom (approximately 60%) is higher for people under 25, whereas it is approximately 50% for people over 25. Younger people also spent a larger percentage of the time that they are at home in the living room on weekends compared to weekdays. However, overall, the percentage of time spent in each of these interior space types have highly similar trends across the age groups and on both weekdays and weekends. It is noted, however, that the amount of time (rather than the percentage of time) spent in theses spaces on weekdays versus weekends is different, as people are away more often on weekdays. 42 Figure 2-6. Spatial location of occupants in residential buildings, including under 25 on (a) under 25 and (b) 25 to 54; (c) 55-64, (d) over 65, and on (1) weekdays and (2) weekends 43 Figure 2-7. Time spent in different indoor spaces on (a) weekdays and (b) weekends by age group 7. Conclusions In this study, typical individual occupancy profiles for U.S. residential buildings are created using ATUS and RECS data and compared with the schedules used in current energy modeling methods and tools. The schedules were then mapped to typical households with multiple people. The spatial distribution of occupants in indoor residential spaces was evaluated based on the time of day and the percentage of time people spent in different rooms. The overall key findings can be summarized as follows: • The typical individual occupancy schedules vary significantly based on the age of the occupants and whether it is a weekday or weekend. • The variations in typical individual occupancy schedules among different age groups are much higher in weekdays compared to that in weekends. For people over 65, the occupancy profile remains similar in both weekdays and weekends. • The occupancy schedule profiles used in the DOE residential Reference Building and Building America (BA) Simulation Protocol overestimate the occupancy from 5:00 to 8:00 am and underestimate occupancy from 7:00 to 10:00 pm compared to the typical individual occupancy schedule developed herein. • Overall, the trends of the typical household occupancy schedule profiles are most similar to those currently used in the DOE Reference Building and Building America (BA) Simulation Protocol for the 3- and 4-person households, however there are larger 44 differences between the currently used schedule and the 1- and 2-members household schedules. • The amount of time that people of different age groups are absent from their home on weekdays and weekends captures the different types of distribution profiles that typical occupants have. Around 42 to 44% of occupants under 55 are absent from their home for 8 to 12 hours periods, whereas, for those 55 to 64, only 33% of occupants are absent for this period of time. For, people 65 and older, this value reduces significantly. • When at home, people spend a majority of their time in the bedroom (54-63% in weekdays and 55-62% in weekends) followed by the living room (19-27% in weekdays and 23-27% in weekends). • Based on the total time spent in different areas of a home, the occupancy profiles are similar on weekdays and weekends. This study provides overall idea of the typical profiles of the U.S. population in the United States and can be used as a part of energy simulation tools to predict the overall building performance. In addition, a better understanding of the spatial location distribution is useful for optimizing the deployment of occupancy sensor systems to detect occupant in a residential building. One limitation of this study is that the ATUS and RECS data have been used in combination. However, these two studies’ data are a result of two different methods of data collection and analysis. The reason the data was merged for use in this study is that ATUS data does not have information on the schedules of all household members, only the schedule for one person in the household. Therefore, an additional dataset is needed to define the other occupant(s)’ schedule(s) in the household and link multiple household members together. This is a limitation of the dataset that could be explored in future studies, such as through field-collected data. Another limitation of this study is that the utilized ATUS data is self-reported; self-reported data can include human error that may influence the results of this work. As a future study, more detailed occupant characteristics should be evaluated, for use in the development of an occupancy simulation tool to generate stochastic annual occupancy schedules for different types of residential buildings. In addition, occupancy-dependent energy end uses, such as appliances and plug loads, can further benefit from and be updated based on the finding of this research, and further analysis of ATUS and other related and complimentary datasets. This is because currently, appliance use profiles in 45 currently-used energy simulation tools ususally follow averaged profiles of use, similar to currently used occupancy simulation methods (Building America 2011). Different occupant centric control strategies can also be developed based on the accurate prediction of occupancy profiles as discussed by Naylor et. al., (2018), Park et. al., (2019) and Shen et. al., (2017). In this study age of the occupants is considered as an influential factor influencing occupancy schedules. Additional influential factors may also be considered to impact occupancy profiles, which can be combined with the findings of this study to improve the accuracy of occupancy schedule prediction. With the spatial distribution of occupants in a home, as well as information on the kinds of activities being conducted, this could lead to better representation of stochastic appliance schedules as well. In addition, if a dataset was available that could be linked with the datasets included herein, this would enable a correlation between the number of occupants and the building characteristics such as the number of bedrooms or floor area, the occupancy profile of different type of residential buildings could be better evaluated. 8. Acknowledgements This project is funded by ARPA-E through the ARPA-E SENSOR program. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of ARPA-E. 46 REFERENCES Advanced Energy Design Guides, 2018. Aerts, D, Minnen, J, Glorieux, I, Wouters, I, and Descamps, F. 2014. “A Method for the Identification and Modelling of Realistic Domestic Occupancy Sequences for Building Energy Demand Simulations and Peer Comparison.” Building and Environment75:67-78. ASHRAE. (1989). 90.1 User’s Manual ANSI/ASHRAE/IESNA Standard 90.1-1989. Atlanta, GA: American Society of Heating, Refrigerating and Air-Conditioning Engineers Balvedi, B. F., Ghisi, E, and Lamberts, R. 2018. “A Review of Occupant Behaviour in Residential Buildings.” Energy and Buildings 174: 495–505. “BEopt Home,” BEopt, accessed October 10, 2019, http://beopt.nrel.gov/. Blanc, M. 2011. INSEE - National Institute of Statistics and Economic Studies - France. http://www.insee.fr/en/default.asp. Blight, Thomas S., and Coley, D. A.. 2013. “Sensitivity Analysis of the Effect of Occupant Behaviour on the Energy Consumption of Passive House Dwellings.” Energy and Buildings 66: 183–92. Cantin, R., Kindinis, A., Michel, P., “New approaches for overcoming the complexity of future buildings impacted by new energy constraints”, Futures, 44 (8) (2012), pp. 735-745 Chiou, Y. S., Carley, K. M, Davidson, C. I., and Johnson, M P.. 2011. “A High Spatial Resolution Residential Energy Model Based on American Time Use Survey Data and the Bootstrap Sampling Method.” Energy and Buildings 43 (12): 3528–38. Christensen, C., Anderson, R., Horowitz, S., Courtney, A., and Spencer, J. BEopt (TM) Software for Building Energy Optimization: Features and Capabilities. United States: N. p., 2006. Web. doi:10.2172/891598. Core Writing Team, Pachauri, R. K., and Meyer, L. A. (eds.). 2002. “IPCC, 2014: Climate Change 2014: Synthesis Report. Contribution of Working Groups I, II and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change.” IPCC 40 (1): 151. https://doi.org/10.1046/j.1365-2559.2002.1340a.x. Deru, M., Field K., Studer D., Benne K., Griffith B., Torcellini P., Liu B., et al. 2011. “U.S. Department of Energy Commercial Reference Building Models of the National Building Stock.” Nrel/Tp-5500-46861, no. February 2011: 1–118. Diao, L, Sun Y, Chen Z, and Chen J. 2017. “Modeling Energy Consumption in Residential Buildings: A Bottom-up Analysis Based on Occupant Behavior Pattern Clustering and Stochastic Simulation.” Energy and Buildings 147: 47–66. https://doi.org/10.1016/j.enbuild.2017.04.072. 47 D’Oca, Simona, and Hong T. 2015. “Occupancy Schedules Learning Process through a Data Mining Framework.” Energy and Buildings 88: 395–408. EIA, Annual Energy Review 2010, Annual report, The World Business Council for Sustainable Development, October 2011. EIA, 2017, International Energy Outlook, US Energy Information Administration EIA. 2015, “A Look at the U.S. Commercial Building Stock: Results from EIA's 2012 Commercial Buildings Energy Consumption Survey (CBECS)”, Release date: March 4, 2015. https://www.eia.gov/consumption/commercial/reports/2012/buildstock/ Fabi, V., Andersen, R.V., Corgnati, S.P., Olesen, B.W., “A methodology for modelling energy- related human behaviour: application to window opening behaviour in residential buildings”, Build Simul J (2013), 10.1007/s12273-013-0119-6 Gershuny, J., Sullivan, O. (2017). United Kingdom Time Use Survey, 2014-2015. Centre for Time Use Research, University of Oxford. [data collection]. UK Data Service. SN: 8128, http://doi.org/10.5255/UKDA-SN-8128-1 Glorieux, I., J. Minnen, T.P. van Tienoven, et al. (2015) Website 'Belgisch tijdsbudgetonderzoek' (www.time-use.be), Onderzoeksgroep TOR Vrije Universiteit Brussel & AD Statistiek – Statistics Belgium, Brussel. Hailemariam, E., Goldstein, R., Attar, R., and Khan., A., 2011. “Real-time occupancy detection using decision trees with multiple sensor types.” In Proceedings of the 2011 Symposium on Simulation for Architecture and Urban Design (SimAUD '11). Society for Computer Simulation International, San Diego, CA, USA, 141-148 Hong T, D'Ocaa S, Turnera, W J. N., C.Taylor-Lange, S, “An ontology to represent energy- related occupant behavior in buildings. Part I: Introduction to the DNAs framework”, Building and Environment, Volume 92, October 2015, Pages 764-777, https://doi.org/10.1016/j.buildenv.2015.02.019 Ipsos-RSL and Office for National Statistics, United Kingdom Time Use Survey [Computer File] (3rd edition), UK Data Archive [distributor] Iwashita, G., Akasaka, H., 1997, “The effects of human behavior on natural ventilation rate and indoor air environment in summer—a field study in southern Japan, Energy” Build., 25 (1997), pp. 195-205, 10.1016/S0378-7788(96)00994-2 Kleiminger, W., Beckel, C., Staake, T., and Santini. S., 2013. “Occupancy Detection from Electricity Consumption Data. In Proceedings of the 5th ACM Workshop on Embedded Systems For Energy-Efficient Buildings” (BuildSys'13). ACM, New York, NY, USA, Article 10 , 8 pages. DOI=http://dx.doi.org/10.1145/2528282.2528295 Li., Z.-J., Jiang, Y., Wei, Q.-P., 2014, “Survey on energy consumption of air conditioning in summer in a residential building in Beijing”, Heat. Vent. Air Cond., 37 (2014), pp. 46-51 48 Liao C., Barooah, P., “An integrated approach to occupancy modeling and estimation in commercial buildings”, Proceedings of the 2010 American Control Conference, 30 June- 2 July 2010, Baltimore, MD, USA, IEEE, DOI: 10.1109/ACC.2010.5531035 López-Rodríguez, M.A., Santiago, I., Trillo-Montero, D., Torriti, J., Moreno-Munoz, A., “Analysis and modeling of active occupancy of the residential sector in Spain: an indicator of residential electricity consumption”, Energy Policy, 62 (2013), pp. 742-751, 10.1016/j.enpol.2013.07.095 McKenna, E., Krawczynski, M., Thomson, M., “Four-state domestic building occupancy model for energy demand simulations”, Energy Build., 96 (2015), pp. 30-39, 10.1016/j.enbuild.2015.03.013. Naylor, S., Gillott, M., & Lau, T.(2018). A review of occupant-centric building control strategies to reduce building energy use. Renewable and Sustainable Energy Reviews, 96, 1-10. Page, J., Robinson, D., Morel, N. and Scartezzini. J. L. 2008. “A Generalised Stochastic Model for the Simulation of Occupant Presence.” Energy and Buildings 40 (2): 83–98. Park, J. Y., Ouf, M. M., Gunay, B., Peng, Y., O'Brien, W., Kjærgaard, M. B., & Nagy, Z. (2019). A critical review of field implementations of occupant-centric building controls. Building and Environment, 106351. Pedersen, T. H., Nielsen, K. U., Petersen, S., “Method for room occupancy detection based on trajectory of indoor climate sensor data”, Building and Environment, Volume 115, April 2017, Pages 147-156, https://doi.org/10.1016/j.buildenv.2017.01.023 Popoola, O. M. 2018. “Computational Intelligence Modelling Based on Variables Interlinked with Behavioral Tendencies for Energy Usage Profile – A Necessity.” Renewable and Sustainable Energy Reviews 82 (May 2017): 60–72.doi.org/10.1016/j.rser.2017.09.020. Richardson, I, Thomson M., and Infield D. 2008. “A High-Resolution Domestic Building Occupancy Model for Energy Demand Simulations.” Energy and Buildings 40 (8): 1560– 66. https://doi.org/10.1016/j.enbuild.2008.02.006. Saha, H., Florita, A. R., Henze, G. P., Sarkar., S., 2019, “Occupancy sensing in buildings: A review of data analytics approaches”, Energy&Buildings188–189 (2019)278–285 Shen, W., Newsham, G., & Gunay, B. (2017). Leveraging existing occupancy-related data for optimal control of commercial office buildings: A review. Advanced Engineering Informatics, 33, 230-242. U.S. Bureau of Labor Statistics, American Time Use Survey, (2009). http:// www.bls.gov/tus/#data (Accessed on 10th June, 2018.). U.S. Energy Information Administration, Residential Energy Consumption Survey, (2009). http://www.eia.gov/consumption/residential/data/2009/ index.cfm?view=microdata (Accessed on 10th June, 2018.). 49 U.S. Energy Information Administration, Residential Energy Consumption Survey, (2015). Vale, Raúl Fraguela, Juan José Lorenzo Castiñeiras, and Lara Varela Garrote. "Escuela, familias y ocio en la conciliación de los tiempos cotidianos de la infancia." Revista de Investigación Educativa 29, no. 2 (2011): 429-446. Vazquez, F. I., and Kastner W. 2011. “Clustering Methods for Occupancy Prediction in Smart Home Control.” Proceedings - ISIE 2011: 2011 IEEE International Symposium on Industrial Electronics, 1321–28. Wagner, A, O’Brien W., and Dong B., "Exploring Occupant Behavior in Buildings." Wagner, A., O’Brien, W., Dong, B., Eds (2018). Wang, W., Chen, J, Song, X, “Modeling and predicting occupancy profile in office space with a Wi-Fi probe-based Dynamic Markov Time-Window Inference approach”, Building and Environment 124 (2017), 130-142 Widén, J., and Wäckelgård E. 2010. “A High-Resolution Stochastic Model of Domestic Activity Patterns and Electricity Demand.” Applied Energy 87 (6): 1880–92. Wilke U, Haldi F, Robinson D.,2011. “A model of occupants’ activities based on time use survey data” Proceed Build Simul 2011 Wilke, U., Haldi, F., Scartezzini, J.-L., Robinson, D. 2013 “A bottom-up stochastic model to predict building occupants’ time-dependent activities”, Build Environment 60 (2013) 254–264, doi:10.1016/j.buildenv.2012.10.021 Wilson, E., Engebrecht Metzger, C., Hendron, R., and Horowitz S.. 2014. 2014 Building America House Simulation Protocols. National Renewable Energy Laboratory. Vol. NREL/TP-55. Xu, X., Chen, C., 2019. “Energy efficiency and energy justice for U.S. low-income households: An analysis of multifaceted challenges and potential.” Energy Policy 128 (2019) 763–774 Yan, D., and Hong, T., 2018. "EBC Annex 66 Final Report-Definition and Simulation of Occupant Behavior in Buildings." Yoshino H, Hong T, Nord N, “IEA EBC annex 53: Total energy use in buildings—Analysis and evaluation methods”, Energy and Buildings, Volume 152, 1 October 2017, 124-136 Zhao, H. X., and Magoulès F. 2012. “A Review on the Prediction of Building Energy Consumption.” Renewable and Sustainable Energy Reviews 16 (6): 3586–92. Zou, H, Hao J, Jianfei Y, Lihua X, and Costas S. "Non-intrusive occupancy sensing in commercial buildings." Energy and Buildings154 (2017): 633-643. 50 APPENDIX Table A1. Indoor location mapping Activity Activity Description Activity Activity Description code code Bedroom 10101 Sleeping 40106 Talking with/listening to nonhh children 10102 Sleeplessness 40199 Caring for and helping nonhh children, n.e.c.* 10199 Sleeping, n.e.c.* 40301 Providing medical care to nonhh children 10401 Personal/Private activities 40399 Activities related to nonhh child's health, n.e.c.* 10499 Personal activities, n.e.c.* 40401 Physical care for nonhh adults 30101 Physical care for hh children 40402 Looking after nonhh adult (as a primary activity) 30102 Reading to/with hh children 40403 Providing medical care to nonhh adult 30103 Playing with hh children, not sports 40499 Caring for nonhh adults, n.e.c.* 30104 Arts and crafts with hh children 40599 Helping nonhh adults, n.e.c.* 30106 Talking with/listening to hh children 49999 Caring for & helping nonhh members, n.e.c.* 30199 Caring for & helping hh children, n.e.c.* 80402 Using in-home health and care services 30301 Providing medical care to hh children 80499 Using medical services, n.e.c.* 30399 Activities related to hh child's health, 90103 Using clothing repair and cleaning n.e.c.* services 30401 Physical care for hh adults 120301 Relaxing, thinking 30402 Looking after hh adult (as a primary 120302 Tobacco and drug use activity) 30403 Providing medical care to hh adult 120312 Reading for personal interest 30499 Caring for household adults, n.e.c.* 120313 Writing for personal interest 30599 Helping household adults, n.e.c.* 130109 Dancing 39999 Caring for & helping hh members, 150103 Reading n.e.c.* 40101 Physical care for nonhh children 150105 Writing 40102 Reading to/with nonhh children 150203 Providing care 40103 Playing with nonhh children, not sports 120312 Reading for personal interest 40104 Arts and crafts with nonhh children 40199 Caring for and helping nonhh children, 80499 Using medical services, n.e.c.* n.e.c.* 40301 Providing medical care to nonhh 90103 Using clothing repair and cleaning children services 40399 Activities related to nonhh child's 120301 Relaxing, thinking health, n.e.c.* 40401 Physical care for nonhh adults 120302 Tobacco and drug use 51 Table A1(cont'd) 40402 Looking after nonhh adult (as a primary 120312 Reading for personal interest activity) 40403 Providing medical care to nonhh adult 120313 Writing for personal interest 40499 Caring for nonhh adults, n.e.c.* 130109 Dancing 40599 Helping nonhh adults, n.e.c.* 150103 Reading 49999 Caring for & helping nonhh members, 150105 Writing n.e.c.* 80402 Using in-home health and care services 150203 Providing care Bathroom 10201 Washing, dressing and grooming oneself 10399 Self care, n.e.c.* 10299 Grooming, n.e.c.* 80501 Using personal care services 10301 Health-related self care 80599 Using personal care services, n.e.c.* Dining room 20104 Storing interior hh items, inc. food 90102 Using meal preparation services 20201 Food and drink preparation 110101 Eating and drinking 20202 Food presentation 110199 Eating and drinking, n.e.c.* 20203 Kitchen and food clean-up 110201 Waiting associated w/eating & drinking 20299 Food & drink prep, presentation, & 110299 Waiting associated with eating & clean-up, n.e.c.* drinking, n.e.c.* 40501 Housework, cooking, & shopping 119999 Eating and drinking, n.e.c.* assistance for nonhh adults 50202 Eating and drinking as part of job 150201 Food preparation, presentation, clean- up Living room 10501 Personal emergencies 120399 Relaxing and leisure, n.e.c.* 10599 Personal care emergencies, n.e.c.* 120501 Waiting assoc. w/socializing & communicating 19999 Personal Care, n.e.c.* 120502 Waiting assoc. w/attending/hosting social events 20101 Interior cleaning 120503 Waiting associated with relaxing/leisure 20103 Sewing, repairing, & maintaining 120599 Waiting associated with socializing, textiles n.e.c.* 20199 Housework, n.e.c.* 129999 Socializing, relaxing, and leisure, n.e.c.* 20301 Interior arrangement, decoration, & 130105 Playing billiards repairs 20302 Building and repairing furniture 130201 Watching aerobics 20303 Heating and cooling 130202 Watching baseball 20399 Interior maintenance, repair, & 130203 Watching basketball decoration, n.e.c.* 20601 Care for animals and pets (not 130204 Watching biking veterinary care) 52 Table A1(cont’d) 20602 Walking / exercising / playing with 130205 Watching billiards animals 20699 Pet and animal care, n.e.c.* 130206 Watching boating 20905 Home security 130207 Watching bowling 20999 Household management, n.e.c.* 130208 Watching climbing, spelunking, caving 29999 Household activities, n.e.c.* 130209 Watching dancing 30105 Playing sports with hh children 130210 Watching equestrian sports 30109 Looking after hh children (as a primary 130211 Watching fencing activity) 30111 Waiting for/with hh children 130212 Watching fishing 30203 Home schooling of hh children 130213 Watching football 30204 Waiting associated with hh children's 130214 Watching golfing education 30303 Waiting associated with hh children's 130215 Watching gymnastics health 30405 Waiting associated with caring for 130216 Watching hockey household adults 30504 Waiting associated with helping hh 130217 Watching martial arts adults 40109 Looking after nonhh children (as 130218 Watching racquet sports primary activity) 40111 Waiting for/with nonhh children 130219 Watching rodeo competitions 40203 Home schooling of nonhh children 130220 Watching rollerblading 40204 Waiting associated with nonhh 130221 Watching rugby children's education 40303 Waiting associated with nonhh 130222 Watching running children's health 40405 Waiting associated with caring for 130223 Watching skiing, ice skating, nonhh adults snowboarding 40503 Animal & pet care assistance for nonhh 130224 Watching soccer adults 40508 Waiting associated with helping nonhh 130225 Watching softball adults 50104 Waiting associated with working 130226 Watching vehicle touring/racing 50201 Socializing, relaxing, and leisure as part 130227 Watching volleyball of job 50205 Waiting associated with work-related 130228 Watching walking activities 50301 Income-generating hobbies, crafts, and 130229 Watching water sports food 50305 Waiting associated with other income- 130230 Watching weightlifting/strength generating activities training 80403 Waiting associated with medical 130231 Watching people working out, services unspecified 80502 Waiting associated w/personal care 130232 Watching wrestling services 53 Table A1(cont’d) 90101 Using interior cleaning services 140103 Waiting associated w/religious & spiritual activities 90104 Waiting associated with using 140105 Religious education activities household services 90199 Using household services, n.e.c.* 150102 Organizing and preparing 90201 Using home 150104 Telephone calls (except hotline maint/repair/décor/construction svcs counseling) 90202 Waiting associated w/ home 150202 Collecting & delivering clothing & main/repair/décor/constr other goods 90299 Using home maint/repair/décor/constr 150204 Teaching, leading, counseling, services, n.e.c.* mentoring 90301 Using pet services 150302 Indoor & outdoor maintenance, repair, & clean-up 90302 Waiting associated with pet services 150399 Indoor & outdoor maintenance, building & clean-up activities, n.e.c.* 90399 Using pet services, n.e.c.* 150401 Performing 90402 Waiting associated with using lawn & 150499 Participating in performance & garden services cultural activities, n.e.c.* 90502 Waiting associated with vehicle main. or 150701 Waiting associated with volunteer repair svcs activities 99999 Using household services, n.e.c.* 150799 Waiting associated with volunteer activities, n.e.c.* 100101 Using police and fire services 150801 Security procedures related to volunteer activities 100102 Using social services 150899 Security procedures related to volunteer activities, n.e.c.* 100304 Waiting associated with using 159999 Volunteer activities, n.e.c.* government services 100305 Waiting associated with civic 160101 Telephone calls to/from family obligations & participation members 100399 Waiting assoc. w/govt svcs or civic 160102 Telephone calls to/from friends, obligations, n.e.c.* neighbors, or acquaintances 120101 Socializing and communicating with 160103 Telephone calls to/from education others services providers 120199 Socializing and communicating, n.e.c.* 160104 Telephone calls to/from salespeople 120201 Attending or hosting 160105 Telephone calls to/from professional parties/receptions/ceremonies or personal care svcs providers 120299 Attending/hosting social events, n.e.c.* 160106 Telephone calls to/from household services providers 120303 Television and movies (not religious) 160107 Telephone calls to/from paid child or adult care providers 120304 Television (religious) 160108 Telephone calls to/from government officials 120305 Listening to the radio 160199 Telephone calls (to or from), n.e.c.* 120306 Listening to/playing music (not radio) 160201 Waiting associated with telephone calls 120307 Playing games 160299 Waiting associated with telephone calls, n.e.c.* 54 Table A1(cont’d) 120311 Hobbies, except arts & crafts and 169999 Telephone calls, n.e.c.* collecting Office room 20901 Financial management 60399 Research/homework n.e.c.* 20902 Household & personal organization and 60401 Administrative activities: class for planning degree, certification, or licensure 20903 HH & personal mail & messages 60402 Administrative activities: class for (except e-mail) personal interest 20904 HH & personal e-mail and messages 60403 Waiting associated w/admin. activities (education) 30108 Organization & planning for hh children 60499 Administrative for education, n.e.c.* 30201 Homework (hh children) 69999 Education, n.e.c.* 30299 Activities related to hh child's education, 70104 Shopping, except groceries, food and n.e.c.* gas 30302 Obtaining medical care for hh children 70105 Waiting associated with shopping 30404 Obtaining medical and care services for 70199 Shopping, n.e.c.* hh adult 30501 Helping hh adults 70201 Comparison shopping 30502 Organization & planning for hh adults 70299 Researching purchases, n.e.c.* 40108 Organization & planning for nonhh 70301 Security procedures rel. to consumer children purchases 40201 Homework (nonhh children) 70399 Security procedures rel. to consumer purchases, n.e.c.* 40299 Activities related to nonhh child's educ., 79999 Consumer purchases, n.e.c.* n.e.c.* 40302 Obtaining medical care for nonhh 80101 Using paid childcare services children 40404 Obtaining medical and care services for 80102 Waiting associated w/purchasing nonhh adult childcare svcs 40505 Financial management assistance for 80199 Using paid childcare services, n.e.c.* nonhh adults 40506 Household management & paperwork 80201 Banking assistance for nonhh adults 50101 Work, main job 80202 Using other financial services 50102 Work, other job(s) 80203 Waiting associated w/banking/financial services 50103 Security procedures related to work 80299 Using financial services and banking, n.e.c.* 50199 Working, n.e.c.* 80301 Using legal services 50204 Security procedures as part of job 80302 Waiting associated with legal services 50299 Work-related activities, n.e.c.* 80399 Using legal services, n.e.c.* 50302 Income-generating performances 80601 Activities rel. to purchasing/selling real estate 50303 Income-generating services 80602 Waiting associated w/purchasing/selling real estate 55 Table A1(cont’d) 50304 Income-generating rental property 80699 Using real estate services, n.e.c.* activities 50399 Other income-generating activities, 80701 Using veterinary services n.e.c.* 50401 Job search activities 80702 Waiting associated with veterinary services 50403 Job interviewing 80799 Using veterinary services, n.e.c.* 50404 Waiting associated with job search or 80801 Security procedures rel. to interview professional/personal svcs. 50405 Security procedures rel. to job 80899 Security procedures rel. to search/interviewing professional/personal svcs n.e.c.* 50499 Job search and Interviewing, n.e.c.* 89999 Professional and personal services, n.e.c.* 59999 Work and work-related activities, n.e.c.* 100103 Obtaining licenses & paying fines, fees, taxes 60101 Taking class for degree, certification, or 100199 Using government services, n.e.c.* licensure 60102 Taking class for personal interest 100401 Security procedures rel. to govt svcs/civic obligations 60103 Waiting associated with taking classes 100499 Security procedures rel. to govt svcs/civic obligations, n.e.c.* 60104 Security procedures rel. to taking classes 109999 Government services, n.e.c.* 60199 Taking class, n.e.c.* 120308 Computer use for leisure (exc. Games) 60201 Extracurricular club activities 150101 Computer use 60203 Extracurricular student government 150106 Fundraising activities 60204 Waiting associated with extracurricular 150199 Administrative & support activities, activities n.e.c.* 60299 Education-related extracurricular 150299 Social service & care activities, activities, n.e.c.* n.e.c.* 60301 Research/homework for class for 150501 Attending meetings, conferences, & degree, certification, or licensure training 60302 Research/homework for class for pers. 150599 Attending meetings, conferences, & interest training, n.e.c.* 60303 Waiting associated with research/homework Garage 20701 Vehicle repair and maintenance (by self) 90501 Using vehicle maintenance or repair services 20799 Vehicles, n.e.c.* 90599 Using vehicle maint. & repair svcs, n.e.c.* 40504 Vehicle & appliance maintenance/repair assistance for nonhh adults ` 20102 Laundry 130124 Running 20801 Appliance, tool, and toy set-up, repair, 130128 Using cardiovascular equipment & maintenance (by self) 56 Table A1(cont’d) 20899 Appliances and tools, n.e.c.* 130131 Walking 50203 Sports and exercise as part of job 130133 Weightlifting/strength training 60202 Extracurricular music & performance 130134 Working out, unspecified activities 120309 Arts and crafts as a hobby 130136 Doing yoga 120310 Collecting as a hobby 130199 Playing sports n.e.c.* 130101 Doing aerobics 140102 Participation in religious practices 130104 Biking 149999 Religious and spiritual activities, n.e.c.* 57 3. CHAPTER 3 – CLUSTER ANALYSIS OF OCCUPANCY SCHEDULES IN RESIDENTIAL BUILDINGS IN THE UNITED STATES Mitra, Debrudra, Yiyi Chu, and Kristen Cetin. "Cluster analysis of occupancy schedules in residential buildings in the United States." Energy and Buildings 236 (2021): 110791. 1. Abstract The energy performance of residential buildings significantly depends on the building occupants’ behavior, which can be highly variable. When the heating, ventilation, and air conditioning (HVAC) system is controlled based on the presence or absence of occupants in a building, occupant behavior is of even further importance to its energy performance. In current practice, building energy simulation tools generally use a single occupancy profile to represent the building’s occupancy schedule, the schedule of which is considered to be the same, regardless of the type of household being modeled. Thus, there is significant potential for improvement to allow for more flexibility and accuracy in calculation of occupancy. The objective of this study is to assess the variations in the typical types of occupancy schedules followed by the U.S. population using cluster analysis. American Time Use Survey data, which statically represents the overall U.S. population’s activities, across 12 years (2006 to 2017), is used. The ATUS data is segregated into smaller groups based on age and weekday/weekend, then divided into activities that are considered “at home” and “away from home”, which are mapped to the presence or non-presence of occupants in the home. Cluster analysis is then used to identify common types of occupancy schedule patterns for each age group. Three main types of patterns are obtained from cluster analysis for each age group, which together represent approximately 88% of people in the United States. The output of the cluster analysis is further analyzed to evaluate the variation in characteristics, including the number of times leaving home, time of day when leaving the home, and the timespan of absence from the home. The results of this study provide detailed insights on how typical occupants in the United States spend their time in residential spaces which can be used to create occupancy profiles for residential buildings. These occupancy profiles could be utilized inform an assessment of the energy use impact of occupancy-based controls of energy consuming systems and technologies. 58 2. Introduction Occupant behavior in buildings has been an emerging area of research in recent years, the study of which has significant benefits to the efficient design of building spaces, heating, ventilating and air conditioning (HVAC) systems, and lighting systems, as well as to the improvement of thermal comfort controls [1]. Six key factors are considered to influence energy consumption in the building sector, including the (a) building envelope, (b) equipment, its (c) operation and maintenance, (d) indoor comfort criteria, (e) occupant behavior and (f) weather conditions [2]. Among these six parameters, (e) occupant behavior, (d) indoor comfort criteria and (c) operations and maintenance generally have a comparatively greater impact on energy performance. Occupants’ behavior and lifestyle choices (e) significantly influence the building indoor environment and overall energy consumption [3, 4]. Lo et al. [5] found that accurate occupant detection can reduce the total energy use in an open space office building by up to 30%. Among climate conditions, housing type and occupant behavior, occupant behavior was found to have the most influence on cooling energy consumption in residential buildings [6]. Variation in occupants’ behavior have also been found to result in an up to 87% change in the air change rate in a residential building which uses both air conditioning and operable windows [7]. A study of 25 households in a residential building in Beijing, China found that HVAC electricity consumption varies from 0 to 14 kWh/m2, due in large part to the variations in how occupants adjust their HVAC system setpoints and controls [8]. An et al. (2018) studied the energy consumption of air conditioning in different rooms in a residential building sector in China, obtaining up to different profiles were obtained based on the occupant’s utilization of the air conditioning system [9]. In another study, Xia et al. (2019) completed cluster analysis of the energy consumption used by the air condition system in residential building in China and showed three different consumption patterns of the air conditioning system, which varied depending on occupant preferences [10]. It was also shown that, compared to commercial buildings, occupant behavior has a more significant impact on residential building energy consumption [11, 12, 13]. This demonstrates the significance of occupancy prediction in residential buildings, as much of energy consumption is driven by occupants’ use of energy-consuming devices. Therefore, in order to evaluate the overall energy consumption in a residential building, accurate prediction of occupancy schedules is needed. 59 The development of occupancy schedules has been an ongoing topic in recent literature; however, most studies focus on assessing the occupancy profile for a single or small subset of buildings [14, 15]. Occupancy prediction based on the data of a single building can predict the occupancy profile for that particular building, however it may not necessarily accurately represent other residential buildings’ occupancy characteristics. Therefore, there is a significant need to evaluate the characteristics of the overall occupancy schedules for the U.S. population, for use in improving how occupancy is represented in building energy models. Currently in building energy simulation models, static occupancy schedules are used, which provide a “typical” occupancy schedule in the United States over a year long period. However, the drawback of this is that if a typical schedule is used throughout the entire simulation period, the daily and hourly variations in occupancy are not represented. In a typical energy model which uses such schedules, there are no situations where no one present in the home (i.e. an occupancy fraction of zero). This becomes problematic when estimating energy and/or demand savings from the use of occupant-centric controls. In addition, there have been very few studies that characterize the different types of occupancy profiles for the U.S. population as a whole. This study seeks to address these two challenges. Several studies in recent literature have focused on estimating the occupancy patterns in buildings. Different approaches have been considered to evaluate occupancy schedules. Statistical methods, data-driven methods and survey-based analyses are among the most commonly used methods. To evaluate the stochastic nature of occupancy prediction, Markov chain methods are among the most commonly used in recent literature. One of the pioneer studies in this field by Page et al. uses a discrete time Markov chain model to evaluate stochastic occupancy profiles [16]. This model works well for single zone, single occupancy scenarios, for shorter spans of absence from the studied building(s). Richardson et al. also studied first order Markov Chain models to evaluate occupancy scenarios in a building space [17]. Adamopoulou et al. developed Markov and semi- Markov models based on the data obtained from camera and motion sensors, to simulate the occupancy for different spaces during different times across a one-day period [18]. Two stochastic models were proposed by Chen et al. for multi-occupant, single zone and multi-occupant, multi- zone scenarios [19]. For the multi-occupant single zone scenario, an inhomogeneous Markov chain method was used for occupancy prediction. Beyond the use of Markov chain models, other methods such as recursive algorithms have also been used for daily occupancy forecasting. 60 However, as mentioned, most studies have focused on a single building or set of buildings, which is difficult to use to generalize for the overall U.S. population [20]. Different data mining algorithms are also used to explore occupancy scenarios. Clustering techniques can be used to identify distinct features among occupancy schedules, allowing for the classification of people into different categories based on their typical schedules. Clustering-based algorithms have also been used in recent literature for electricity consumption data to estimate occupancy patterns in buildings [21]. Buttitta et al. used cluster analysis to identify groups of households with similar types of occupancy profiles for five regions in the UK [22]. Zhao et al. used C4.5 Decision Tree and Support Vector Machine (SVM) methods based on data from Bluetooth-connected devices to predict occupancy schedules [23]. In other research, neural network methods, including Artificial Neural Networks, Convolutional Neural Networks, and Recurrent Neural Networks are also used to develop representative occupancy profiles [24, 25, 26]. However, most studies focus on commercial buildings, with evaluation of performance limited to the specific location of testing only. Several survey-based studies have also focused on occupancy schedule prediction in residential buildings. Balvedi et al. developed a 24-hour and 48-hour schedule for weekdays and weekends, respectively, based on interviews with occupants [27]. Similarly, Carpino et al. conducted a questionnaire-based study across 80 families in Italy for 2 weeks in 2017 to evaluate the percentage of time people spent in their home [28]. Occupancy profiles and their interaction with energy- consuming appliances for residential buildings has also been studied in China based on the completion of a questionnaire and subsequent interviews of occupants [29]. These survey-based studies provide insights on occupant behavior and their interactions with building systems. However, conducting similar studies to represent the overall U.S. population would be expensive and time intensive. To study occupant behavior for the U.S. population, it is important to use representative data. However, it is challenging to create a dataset to represent the typical characteristics of occupant behaviors [30]. Time Use Survey data can fulfil this data requirement, as it statistically represent the overall population of a country. Several countries, including China, Japan, Belgium, the U.S., etc… publish their time use survey data which contains detail about their respective populations’ 61 characteristics and their activities throughout the day [31, 32]. Aerts et al. used clustering methods with the Belgian Time Use survey for 2005 data to identify the typical occupancy profiles in residential buildings [33]. Occupancy profiles and their interaction with the building energy systems were evaluated using the UK Time Use Survey data [34]. UK Time Use Survey data was also used to predict the location and activities of occupants using time-inhomogeneous first order Markov Chain model [35]. Similarly, French Time Use survey was used to characterize occupants based on their activity profiles [12]. Spanish Time Use survey data from 2009-2010 was used to predict the occupancy and pattern of energy consumption in the residential buildings [36]. ATUS data from 2006 was used to characterize different activity patterns of occupants by [37]. However, the variation in the occupant profiles for households with different numbers of members is not evaluated in these recent studies. In addition, additional study of data with more recent datasets is also needed for the improved characterization of occupancy profiles. In summary, occupant behavior is among the most important parameters in predicting energy consumption in residential buildings. Detailed study is required to analyze characteristics of occupancy schedules to develop a better understanding of the variations in occupancy schedules. In order to address the previously discussed challenges, this paper uses a clustering algorithm to analyze the different type of typical schedules occupants follow in residential building in the United States. To achieve this, American Time Use Survey (ATUS) data from 12 years, 2006 to 2017 [38] was used. Factors influencing occupancy schedules, including age of the occupants and weekday or weekend, are also considered in this analysis. The results of this work provide unique insights using data across a longer time span as compared to previous studies, and the most recent data available. In addition, unlike previous analyses, this study also analyzes the characteristics of each of the profiles. This provides a clearer picture of occupancy, for use in assessing the potential energy and demand savings from occupancy-centric control in residential buildings. This paper is organized in four sections. First the dataset characteristics are discussed, followed by data processing and the clustering methodology used for overall data analysis. Next the results are discussed, comparing variations in results according to age, and day of the week. Different characteristics of the cluster profiles are also analyzed in this study. This is followed by the conclusions, limitations, and ongoing and future work. The results of this study can be used to 62 create an occupancy simulator for the residential buildings in the United States to predict schedules based on known occupant characteristics. 3. Datasets The American Time Use Survey (ATUS) [37] is an annual survey conducted by the United States Bureau of Labor Statistics. The objective of this survey is to collect information for overall U.S. population, on the activities people complete throughout the day. This information is collected through a combination of email, telephone and in-person interviews. Participants are asked to document their activities, and where the activities take place, and whenever there is any change in activities. In this way, people do not need to update their activities and associated locations at specific time intervals as is done for time use survey data collection in many European countries [39]. ATUS data is collected over a 24-hour time period, starting from 4 am and ending at 4 am of the following day. An example of activity information in ATUS data is shown in Table 1, where in column 1, the person is identified with a unique ID. In total, 19 activities were completed by the example participant for that day, each of which has a starting and ending time. The codes in the last three columns represent the specific type of activity being conducted, divided into three tiers. The first tier includes 18 major categories, which then, using the subcategories of activities in tier 2 and 3, enables a maximum 470 types of activities. However, ATUS only collects the information about the primary activities completed by the participants. If the participant is doing more than one activity at once, such as doing laundry and watching television, only the primary activity is recorded. The TEWHERE and TUACTDUR24 variables represent the location and the duration of the activities in minutes, respectively. 63 Table 3-1. Example activity data from the ATUS survey TUCASEID ACTIVITY_ TEWHERE TUACTDUR START STOP TIER1 TIER2 TIER3 N 24 TIM TIME CODE CODE CODE 2017010117002 1 -1 90 4:00:00 5:30:00 1 1 1 2017010117002 2 -1 30 5:30:00 6:00:00 1 2 1 2017010117002 3 12 15 6:00:00 6:15:00 18 5 1 2017010117002 4 2 15 6:15:00 6:30:00 5 1 4 2017010117002 5 2 150 6:30:00 9:00:00 5 1 1 2017010117002 6 2 30 9:00:00 9:30:00 11 1 1 TUCASEID TUACTIVITY_N TEWHERE TUACTDUR24 TUSTARTTIM TUSTOPTIME TUTIER1CODE TUTIER2CODE TUTIER3CODE 2017010117002 7 2 300 9:30:00 14:30:00 5 1 1 20170101170002 1 -1 90 4:00:00 5:30:00 1 1 1 2017010117002 20170101170002 2 8 -1 2 30 305:30:00 14:30:00 6:00:00 15:00:001 12 2 3 1 1 2017010117002 20170101170002 3 9 12 12 15 156:00:00 15:00:00 6:15:00 15:15:00 18 18 5 3 1 1 20170101170002 4 2 15 6:15:00 6:30:00 5 1 4 2017010117002 20170101170002 5 10 2 3 150 3 6:30:00 15:15:00 9:00:00 15:18:005 3 1 1 1 12 2017010117002 20170101170002 6 11 2 12 30 3 9:00:00 15:18:00 9:30:00 15:21:00 11 18 1 3 1 1 20170101170002 7 2 300 9:30:00 14:30:00 5 1 1 2017010117002 12 1 45 15:21:00 16:06:00 2 2 1 20170101170002 8 2 30 14:30:00 15:00:00 12 3 1 2017010117002 20170101170002 9 13 12 1 15 3015:00:00 16:06:00 15:15:00 16:36:00 18 11 3 1 1 1 20170101170002 10 3 3 15:15:00 15:18:00 3 1 12 2017010117002 14 1 10 16:36:00 16:46:00 2 2 3 20170101170002 11 12 3 15:18:00 15:21:00 18 3 1 2017010117002 20170101170002 12 15 1 -1 45 1515:21:00 16:46:00 16:06:00 17:01:002 1 2 2 1 1 20170101170002 13 1 30 16:06:00 16:36:00 11 1 1 2017010117002 16 1 14 17:01:00 17:15:00 12 3 3 20170101170002 14 1 10 16:36:00 16:46:00 2 2 3 2017010117002 20170101170002 15 17 -1 -1 15 40516:46:00 17:15:00 17:01:00 0:00:001 1 2 1 1 1 20170101170002 2017010117002 16 18 1 -1 14 7 17:01:00 17:15:00 0:00:00 12 0:07:00 1 3 2 3 1 20170101170002 17 -1 405 17:15:00 0:00:00 1 1 1 2017010117002 20170101170002 18 19 -1 -1 7 2330:00:00 0:07:00 0:07:00 5:30:001 1 2 1 1 1 2017010117012 20170101170002 19 1 -1 -1 233 1800:07:00 4:00:00 5:30:00 7:00:001 1 1 1 1 1 20170101170012 1 -1 180 4:00:00 7:00:00 1 1 1 2017010117022 20170101170012 2 2 1 1 60 607:00:00 7:00:00 8:00:00 8:00:002 2 2 2 1 1 *Note: TUCASEID : Participants identifier; TUACTIVITY_N : Number of activities of a participants; TEWHERE : Location of the participants; TUACTDUR24 : Duration of each activities; TUSTARTTIME : Starting time of an activity; TUSTOPTIME : Stopping time of the activity; TUTIER1CODE : Activity code (1st tier); TUTIER2CODE : Activity code (2nd tier); TUTIER3CODE : Activity code (3rd tier); People of all ages participate in this study. However, if the individuals are under 15 or over 75, their ages are denoted as 15 and 75, respectively, in the data. This survey was first published in 2003, then published annually after that. To represent the overall U.S. population statistically, a weightage factor is used, to reduce the biases in the dataset due to different response rates of people across different subpopulations and days of the week. We note that even though the survey has been conducted since 2003, the method to calculate the weightage factor was modified slightly in 2006 and thus only data from 2006 to 2017, which is the most recent available, are used herein. The total number of people interviewed each year, with the number of data obtained for both weekdays and weekends, are shown in Table 2. 64 Table 3-2. Amount of data collected in the ATUS for each year of study for both weekdays and weekends Year Number of Participants surveyed Weekdays Weekends 2006 12943 6486 6457 2007 12248 6080 6168 2008 12723 6202 6521 2009 13133 6554 6579 2010 13260 6591 6669 2011 12479 6304 6175 2012 12443 6108 6335 2013 11385 5702 5683 2014 11592 5825 5767 2015 10905 5475 5430 2016 10493 5327 5166 2017 10223 5079 5144 4. Methodology Activity data was first extracted from the ATUS dataset. The entire dataset was divided into several groups based on the classifiers that were selected based on previous studies, including the age of the participants and whether the day of study was a weekday or weekend [40]. Age was divided into seven categories, including under 25, 25 to 34, 35 to 44, 45 to 54, 55 to 64, 65 to 74 and over 75 years of age, based on categories of age ranges used in the Residential Energy Consumption Survey data [41]. The data was also divided into weekday and weekend data. This results in a total of 14 subsets of data utilized for analysis. Using this data, the variation in the duration of different activities for participants in different age groups, and the location of the activity, are analyzed in this study. For some activities, the location column explicitly provides information on whether the participants are in their home or not, while for other activities such as sleeping, grooming and personal activities, specific locations were not provided. In these situations, it was assumed that these activities occurred in a residential building. For all activities that were considered to be at home, an occupancy fraction of 1 was assigned for the duration of that activity; for all activities that were considered to be elsewhere (i.e. not at home), an occupancy fraction of 0 was assigned. Given there was a significant amount of variation in activities and resulting occupancy patterns 65 among each of these subsets of data, cluster analysis was then used to identify typical patterns of occupancy amongst each subset. Cluster analysis is a popular technique in unsupervised learning, where the objective is to group the studied data in different clusters according to the similarity of data’s characteristics. Such clustering methods are used in recent literature to analyze static or time series data [42, 43]. Specifically, for occupancy schedules, Diao et. al. evaluated 10 different occupant schedule profiles for New York residents [44]. Clustering methods are also used for the study of electricity/energy consumption patterns [45, 46, 47]. Such clustering methods were used in this study to identify the specific temporal characteristics of the resulting occupancy schedules. In this study, the k-means clustering method was applied to each subset of data to evaluate the types of occupancy scenarios present in the ATUS-derived data. The k-means method is a centroid based clustering method where the number k indicates the number of clusters. The algorithm attempts to find k number of centroids of data to subdivide the entire dataset [48]. To implement the cluster analysis, the NbClust package was used in R [49]. The objective of this clustering algorithm is to put objects within a same cluster which are similar to each other and to separate it from the objects which are not so similar. Different number of clusters, from 2 to 6 were evaluated for this study. The effectiveness of the cluster analysis was evaluated using silhouette width (Si), using the “cluster” library in R [50, 51]. The silhouette width analysis is an unsupervised method which evaluates the dissimilarity within the clustered data and among other clusters. The goodness of fit of the cluster results was analyzed using the average of the Si for each cluster. The Si value (Equation 1) evaluates the quality of the clusters by determining whether or not the data lies in the proper cluster [52]. Si is defined as &(!))*(!) 𝑆𝑖(𝑖) = +,- (*(!),&(!)) (1) Where, a(i) and b(i) is the average distance of i from the specific cluster and to the other clusters [53]. An 𝑆𝑖 value of 1 represents data that is “well classified”; a negative 𝑆𝑖 indicates the data has been classified in the wrong cluster. The higher the average of the 𝑆𝑖 values, the better the goodness of fit for the clusters. The cluster analysis supports a better understanding of the specific characteristics of unique patterns of occupancy schedules in the data. The percentage of people in 66 each of the age groups who follow these clustered schedules was calculated. The profiles were then analyzed to evaluate significant characteristics of the profiles, which consist of the number of times people leave their home and when and for how long they are not present in their home. 5. Results 5.1. Activity variation Total duration of different activities within each age group were evaluated and compared (Figure 1). The most common activities in Figure 1 include ‘personal care including sleeping’, ‘household activities’, ‘caring and helping household members’, ‘work related activities’, ‘education’, ‘eating and drinking’, ‘socializing, relaxing and leisure’, ‘traveling’ and ‘other activities’. The ‘other activities’ category consists of activities such as ‘government services & civic obligations’, ‘religious and spiritual activities’, ‘volunteer activities’, ‘consumer purchases’, ‘professional and personal care services’, ‘household services’, ‘caring for and helping non-household members, ‘telephone calls’ and ‘sports, exercise and recreation’. Among those studied, people under the age of 25 spent the most time doing ‘personal care including sleeping’ activities. The amount of time spent doing this activity decreases slightly with increasing age, until the 35 to 44 age category, then increase with increasing age. The percentage time spent doing ‘household activities’ similarly increases with age until the 35 to 44 age category, then remains near constant for the remaining older age groups. ‘Caring for & helping household member’ activities are the most common for the age group of 25 to 34, followed by the 35 to 44 age group. ‘Socializing, relaxing and leisure’ activities decrease with an increase in age until ages 35 to 44; after this age, the percentage increases with age. This activity is mostly common for the older age group over 65 years. These activity variations with respect to age group portray a broad overview about how people spent their time in a typical 24-hour time period. 67 45 40 45 35 40 30 35 Percentage distribution Percentage distribution 25 30 20 25 15 20 10 15 5 10 0 5 Personal care including Work & Work-Related Education EatingEating and Drinking Personal care including Socializing, Relaxing, and and Traveling Household activities CaringCaring For & For & Helping Other activities Helping Work & Work-Related Household activities Education Socializing, Relaxing, Traveling and Drinking Other activities 0 Household Household Members Members Activities sleeping sleeping Activities Below 25 25 to 34 35 to 44 45 to 54 55 to 64 LeisureLeisure 65 to 74 Over 75 Below 25 25 to 34 35 to 44 45 to 54 55 to 64 65 to 74 Over 75 Figure 3-1. Percentage of time spend doing various activities by age group 5.2. Cluster Analysis The result of cluster analysis across the studied age groups can be characterized in three major types of daily profiles irrespective of age group, and weekday/weekend, as shown in Figure 2. The vertical axis represents the centroid of each of the clusters which is the average of the occupancy profiles of all people belonging to that cluster with respect to time of day (0-24 hours). An occupancy fraction of 1 indicates that, across the cluster, all people were at home at a particular time, where a 0 indicated that no one in the cluster was present in their home. A value in between represents the fraction of studied people considered to be at home. It is important to note in this portion of the analysis, the occupancy profiles developed are for individuals, not entire households of multiple people. This is further discussed in later sections. In Figure 2.a, the occupancy fraction value remains close to 1 throughout the day which implies that the people in this group remain in their home for most of the day. During the midday period, the occupancy fraction value drops slightly to around 0.8, indicating that some people leave home for a period of time, perhaps for lunch or other errands. This profile is thus named the stay-home profile in the following sections. In the overall dataset, this profile represents 71% of people across both weekdays and weekends, including approximately 57% of people on weekdays and 90% on weekends. Figure 2.b represents the group of people who are typically present in their home during 68 the night but leave home, on average, around 8:00 am, most likely for work or school. The occupancy fraction remains close to 0 until approximately 4:00 pm, then increases to close to 1 in the evening. This profile represents the occupants who follows an away-from-home schedule during daytime. Thus, this profile is called the day-work profile. This profile represents 26% of all people, including 41% on weekdays, and 7% on weekends. The final profile is opposite of the day- work profile, where the occupancy fraction is 0 or close to 0 during the night, and highest during the daytime. This profile is designated as night-work as it represents people who work during the night; overall approximately 3% of people across both weekdays and weekends follow this schedule. Figure 2.d represents the variation between the schedule used for the residential DOE Reference Building [54] and the three profiles developed in this study. The day-work profile follows a similar trend to the reference schedule but varies in magnitude. The other two profiles differ significantly compared to the reference building profile. 1 1 1 Occupancy fraction 0.8 0.8 0.8 0.6 0.6 0.6 0.4 0.4 0.4 0.2 0.2 0.2 0 0 0 0: 2: 4: 6: 00 00 00 00 0: 2: 4: 00 00 00 0: 2: 4: 00 00 00 8: 10 0 :0 12 0 :0 14 0 :00 6: 8: 10 0 :0 12 0 :000 0 6: 8: 10 0 :0 12 0 :000 0 16 0 :0 18 0 :0 20 0 :0 22 0 14 0 :0 16 0 :0 18 0 :0 20 0 14 0 :0 16 0 :0 18 0 :0 20 0 :0 0 :0 22 0 :0 0 :0 22 0 :0 0 Time of day (a) (b) (c) 1.0 Occupancy fraction 0.8 0.6 0.4 0.2 0.0 0: 1: 3: 4: 6: 7:00 30 00 30 00 9:30 0 10 0 :3 12 0 :0 13 0 :3 15 0 :0 16 0 :3 18 0 :0 19 0 :3 21 0 :0 22 0 :3 0 Time of day Stay home Day work Night work Reference (d) Figure 3-2. Residential occupancy profiles of individuals following (a) stay-home, (b) day-work, (c) night-work schedules and (d) comparison of DOE Reference Building [54] occupancy schedule with the three obtained schedule types 69 The average weekday and weekend profiles of people for each of the age groups are also shown in Figure 3. As seen in Figure 3, even though the three specified cluster types can generally describe the profiles of people across all age groups, the profiles vary somewhat by age. Detailed values of the occupancy fraction for different profiles for both weekdays and weekends are shown in Appendix. Even though most people can be generally clustered into these three profile types, the percent of people who follow these schedules varies significantly by age group and weekdays versus weekends, as shown in Table 3. As seen in the Table, for the developed clusters, for occupants under 25, approximately 31% generally follow the stay-home profile during day on weekdays whereas nearly 95% follow a stay-home profile on weekends. Approximately 65% of people in this age group follow the day-work schedule on weekdays, whereas the remaining portion work at night. For all age groups, up to age 54, this distribution of percentage of people following different schedules is similar. Interestingly, the 25 to 34 age group, while somewhat similar, on weekdays approximately 50% stay at home, 45% follow the day-work schedule and the remaining 5% are not present in home during the night. In other words, a higher percentage of this age group stays at home for the majority of the time on weekdays, compared to the under 35, and 35 to 44 and 45 to 54 age ranges, where the larger portion are away from home. This may be because these people spent a larger amount of time doing activities such as ‘Helping and Caring for Household Members’ which most likely involves caring for children. Over the age of 55, the percentage of people who follow the stay-home schedules increases, and the day-work and night-work schedule percentages decrease. The distribution of schedules is similar for the 65 to 74 and over 75 age groups, as well as for the 35 to 44 and 45 to 54 age groups. This is likely due to people retiring from the workforce as they age. The 35 to 44 and 45 to 54 age typically include people who are still working, whereas the 65 to 74 and over 75 age groups are more likely mostly retired. Those in the 55 to 64 age group are near retirement age or have just reached it, thus there is likely a mix of working and retired people in this group. We also note that these percentage values represent the result of the cluster analysis. Most profiles fit well within the developed clusters, however there were a few that did not. These results are discussed in the later section of this paper. 70 Occupancy fraction Occupancy fraction Occupancy fraction Occupancy fraction Occupancy fraction Occupancy fraction Occupancy fraction 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0:00 0:00 0:00 0:00 0:00 0:00 0:00 1:00 1:00 1:00 1:00 1:00 1:00 1:00 2:00 2:00 2:00 2:00 2:00 2:00 2:00 3:00 3:00 3:00 3:00 3:00 3:00 3:00 4:00 4:00 4:00 4:00 4:00 4:00 4:00 5:00 5:00 5:00 5:00 5:00 5:00 5:00 6:00 6:00 6:00 6:00 6:00 6:00 6:00 7:00 7:00 7:00 7:00 7:00 7:00 7:00 8:00 8:00 8:00 8:00 8:00 8:00 Weekday 9:00 9:00 Weekday 9:00 Weekday 9:00 Weekday 9:00 9:00 8:00 10:00 10:00 10:00 10:00 10:00 10:00 Weekday 9:00 11:00 11:00 of day 11:00 11:00 Time of day 11:00 11:00 10:00 11:00 of day Time of day 12:00 12:00 12:00 12:00 12:00 12:00 13:00 (a7) 13:00 (a6) 13:00 13:00 (a4) 13:00 (a3) 13:00 12:00 Time of day Time of day (a5) (a2) (a1) Stay home Time65-74 Stay home Time55-64 Stay home 45-54 Time of day 13:00 Stay home 35-44 14:00 14:00 14:00 14:00 14:00 14:00 Stay home over 75 Stay home 25-34 Stay home below 25 15:00 15:00 15:00 15:00 15:00 15:00 14:00 Weekend 15:00 Weekday Weekend Weekend Weekend Weekend Weekday Weekend 16:00 16:00 16:00 16:00 16:00 16:00 Weekend 17:00 17:00 17:00 17:00 17:00 17:00 16:00 18:00 18:00 18:00 18:00 18:00 18:00 17:00 19:00 19:00 19:00 19:00 19:00 19:00 18:00 20:00 20:00 20:00 20:00 20:00 20:00 19:00 21:00 21:00 21:00 21:00 21:00 21:00 20:00 22:00 22:00 22:00 22:00 22:00 22:00 21:00 23:00 23:00 23:00 23:00 23:00 23:00 22:00 Figure 3-3. Average occupancy profiles for individuals ages (a) under 25, (b) 25-34, (c) 35-44, 23:00 Occupancy fraction Occupancy fraction Occupancy fraction Occupancy fraction Occupancy fraction Occupancy fraction Occupancy fraction 0.2 0.2 0.4 Occupancy fraction 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0:00 0.0 0.0 0.4 0.6 0.8 1.0 0:00 0:00 0:00 0:00 0:00 0:00 1:00 0:00 1:00 1:00 1:00 1:00 1:00 1:00 2:00 1:00 2:00 2:00 2:00 2:00 2:00 2:00 3:00 2:00 3:00 3:00 3:00 3:00 3:00 3:00 4:00 3:00 4:00 4:00 4:00 4:00 4:00 4:00 5:00 4:00 5:00 5:00 5:00 5:00 5:00 5:00 6:00 5:00 6:00 6:00 6:00 6:00 6:00 6:00 7:00 6:00 7:00 7:00 7:00 7:00 7:00 7:00 8:00 7:00 8:00 8:00 8:00 8:00 8:00 8:00 Weekday 9:00 8:00 9:00 Weekday 9:00 Weekday 9:00 Weekday 9:00 Weekday 9:00 Weekday 9:00 Weekday 10:00 9:00 Day work over 75 10:00 10:00 10:00 10:00 10:00 10:00 71 11:00 10:00 Weekday Weekend 11:00 11:00 Time of day 11:00 11:00 11:00 Time of day 11:00 Time of day 12:00 12:00 12:00 12:00 12:00 12:00 Time of day 12:00 11:00 (b7) (b6) (b5) (b3) (b1) Day work 65-74 Time of day 13:00 13:00 13:00 (b4) Day work 45-54 13:00 13:00 (b2) 13:00 13:00 12:00 Time of day Time of day Day work 55-64 Time of day Day work 35-44 Day work 25-34 Day work below 25 14:00 13:00 14:00 14:00 14:00 14:00 14:00 14:00 14:00 15:00 Night work over 75 15:00 15:00 15:00 15:00 15:00 15:00 Weekend Weekend Weekend Weekend Weekend 15:00 16:00 16:00 16:00 16:00 16:00 16:00 Weekend 16:00 16:00 17:00 17:00 17:00 17:00 17:00 17:00 17:00 17:00 18:00 18:00 18:00 18:00 18:00 18:00 18:00 19:00 19:00 19:00 19:00 19:00 Weekend 18:00 19:00 19:00 19:00 20:00 20:00 20:00 20:00 20:00 20:00 20:00 20:00 21:00 21:00 21:00 21:00 21:00 21:00 21:00 21:00 22:00 22:00 22:00 22:00 22:00 22:00 22:00 23:00 23:00 23:00 23:00 23:00 (d) 45-54, (e) 55-64, (f) 65-74, (g) over 75 for (1) stay-home, (2) day-work and (3) night-work 22:00 23:00 23:00 23:00 Occupancy fraction Occupancy fraction Occupancy fraction Occupancy fraction Occupancy fraction Occupancy fraction Occupancy fraction 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0:00 0:00 0:00 0:00 0:00 0:00 0:00 1:00 1:00 1:00 1:00 1:00 1:00 1:00 2:00 2:00 2:00 2:00 2:00 2:00 2:00 3:00 3:00 3:00 3:00 3:00 3:00 3:00 4:00 4:00 4:00 4:00 4:00 4:00 4:00 5:00 5:00 5:00 5:00 5:00 5:00 5:00 6:00 6:00 6:00 6:00 6:00 6:00 6:00 7:00 7:00 7:00 7:00 7:00 7:00 7:00 8:00 8:00 8:00 8:00 8:00 8:00 8:00 Weekday 9:00 Weekday 9:00 Weekday 9:00 Weekday 9:00 9:00 Weekday 9:00 Weekday 9:00 10:00 10:00 10:00 10:00 10:00 10:00 10:00 11:00 11:00 11:00 11:00 11:00 of day 11:00 11:00 (c7) 12:00 12:00 (c5) 12:00 (c4) 12:00 12:00 12:00 (c6) 12:00 of day (c3) of day 13:00 13:00 13:00 13:00 (c2) 13:00 (c1) 13:00 13:00 Night work 45-54 Night work 25-34 Time of day Night work Time65-74 Night work Time55-64 Night work Time35-44 Time of day Time of day Time of day Night work over 75 14:00 14:00 14:00 14:00 14:00 14:00 14:00 Night work below 25 15:00 15:00 15:00 15:00 15:00 15:00 15:00 Weekend Weekend 16:00 Weekend 16:00 Weekend 16:00 Weekend 16:00 Weekend 16:00 Weekday Weekend 16:00 16:00 17:00 17:00 17:00 17:00 17:00 17:00 17:00 18:00 18:00 18:00 18:00 18:00 18:00 18:00 19:00 19:00 19:00 19:00 19:00 19:00 19:00 20:00 20:00 20:00 20:00 20:00 20:00 20:00 21:00 21:00 21:00 21:00 21:00 21:00 21:00 22:00 22:00 22:00 22:00 22:00 22:00 22:00 23:00 23:00 23:00 23:00 23:00 23:00 23:00 schedules on weekdays and weekends Table 3-3. Percent (%) of people in each age group cluster following the occupancy schedule types on both weekdays and weekends Age range of occupants Day type Cluster type Under 25 25-34 35-44 45-54 55-64 65-74 Over 75 Stay-home 53 50 37 31 69 85 81 Weekday Day-work 43 45 61 66 29 13 18 Night-work 4 5 2 3 2 2 1 Stay-home 95 92 83 93 89 91 95 Weekend Day-work 2 4 14 4 9 8 4 Night-work 3 4 3 3 2 1 1 For the weekends, regardless of age, generally it can be observed that the large majority of people follow the weekend stay-home profiles. However, it is important to note that both the stay-home and day-work profiles for weekends differ from those of the weekdays. As shown in Figure 3, for the stay-home profiles on weekends, the occupancy fraction drops to near 0.6 midday, as compared to 0.8 midday on weekdays. The day-work profile for weekends varies compared to weekdays in terms of the duration that people are away from home. On weekdays, the occupancy fraction remains near 0 for approximately 8 hours from 8:00 am to 4:00 pm, whereas on weekends, an occupancy fraction near 0 occurs for only a short amount of time, from 3 to 6 hours depending on the age group. In addition, the rate of change in the occupancy fraction values on weekends are smaller compared to on weekdays. We also note that for the night-work schedules on both weekdays and weekends, these represent a very small percent of the studied population (see Table 3), ranging from approximately 1% to 5% of each cluster, thus this is likely the cause for the significant increases and decreases that occur in the profiles over a short period of time. Additional supplementary data could be used to further study and verify these profiles. To better quantify the variation across each of the clusters of occupancy profiles, the maximum, minimum and average occupancy fraction for different occupancy schedules across all age groups are shown in Figure 4. For the stay-home profile, the variation for both weekdays and weekends are minimal during the nighttime period and at the start of the day, whereas during the daytime, the variation in the occupancy fraction between the maximum and minimum is near 0.2. Similar variations can be seen for day-work profile. During weekdays, in the daytime the occupancy profiles follow similar pattern; the variation between the minimum and the maximum occupancy fraction during the daytime is near 0.2. The variations in the day-work profile on weekends are higher compared to weekdays. Both at the start and end of the day, the variation between the 72 minimum and maximum occupancy fractions are near 0.2 whereas during daytime, it is near 0.4 which is much higher compared to the variation values on weekdays. The pattern in the maximum and the minimum occupancy fraction also varies slightly during the evening on weekends. These larger variations on weekends may be because this type of schedule is less common on weekends, and perhaps more sporadic for those following such a schedule. The variation in the occupancy fraction for night-work profiles for both weekdays and weekends are similar, where the difference between the maximum and the minimum value is near 0.4. We note the sharp jump in occupancy around approximately 4 am in the night-work profile; this is likely due to limited data for people following this profile, combined with the data being collected from ATUS starting at 4:00 am, which is likely introducing some error into this result. Stay homestay Weekday below 25 home Weekday daywork Weekday nightwork 1.0 1.0 1.0 1.0 Occupancy fraction 0.8 0.8 0.8 0.8 0.6 0.6 0.6 0.6 0.4 (a2) 0.4 (a1) 0.4 0.4 (a3) 0.2 0.2 0.2 0.2 Weekend nightwork 0.0 0.0 0.0 0.0 1.0 0:0 0:00 0 1:1:00 30 3:2:00 00 4:3:00 30 6:4:00 00 5:00 7:6:00 30 9:7:00 00 108:00 :30 129:00 :0 10:00 13 0 11:00 :30 112:00 5:0 113:00 6:3 114:00 8:0 15:00 19 0 16:00 :30 217:00 1:0 0 0 0:0 1:3 3:0 0 0 218:000 0 Stay home below 25 2:3 19:00 20:00 21:00 22:00 23:00 0:0 1:3 3:0 0 0 00 4:3 6:0 7:3 9:0 10 0 0 0 0 4:3 0 :3 12 0 Weekend nightwork 6:0 7:3 9:0 0 0 :0 13 0 :3 Weekend Time stayhome Weekend daywork 10 0 :3 12 0 :0 13 0 :3 15 0 :0 16 0 :3 18 0 15 0 :0 16 0 :3 18 0 :0 19 0 :3 0.8 :0 19 0 :3 21 0 :0 of day 21 0 :0 22 0 :30 22 0 :30 1.0 1.0 Maximum Average Minimum 1.0 Maximum Average Minimum 1.0 Maximum Average Minimum Weekday Weekend Occupancy fraction 0.8 0.6 0.8 0.8 0.8 0.6 0.6 0.4 0.6 0.6 0.4 (b2) 0.4 (b1) 0.4 0.4 0.2 0.2 (b3) 0.2 0.2 0.2 0.0 0.0 0.0 0.0 13 0:0 1:31:00 3:02:00 4:33:00 6:04:00 7:36:00 9:07:00 10 8:00 :30 12 9:00 0:00 0 0 0 0 5:00 0 0 0 :10:00 00 :11:00 30 0.0 15 12:00 :00 22 0 16 13:00 :30 18 14:00 :15:00 00 19 16:00 :30 21 17:00 :00 22 18:00 :19:00 30 020:00 :00 0:0 1:3 3:0 0 0 0 121:00 :30 22:00 03:23:00 :000 0 4:3 6:0 7:3 0 0 :30 14::3 36:0 0 70 :0 4:300 6::030 79::30 109:000 30 0 9:0 10 0 :3 12 0 :0 13 0 :3 15 0 :0 0 : 1102:300 30 16 0 Time of day 12 :0 113 00 : 3::3 155:300 6:3 0 :0 116 000 18 :30 118:000 9 :0 :3 18 0 :0 19 0 :3 21 0 :0 22 0 :30 1219:300 ::03 2221:300 :00 MaximumWeekdayAverage WeekendMinimum Maximum Average Minimum Maximum Average Minimum Maximum Average Minimum Figure 3-4. Maximum, minimum, and average occupancy fraction for all individuals for (a) stay- home, (b) day-work and (c) night-work profiles on both (1) weekdays and (2) weekends 5.3. Cluster Analysis Accuracy The effectiveness of the cluster analysis is calculated using the Si value for different age groups and for weekday and weekends, as shown in Figure 5. Each of the subfigures represents the Si value for each of the occupancy profiles. The majority of the Si values are positive, and a smaller number are negative. The negative values (Figure 6) indicate the percentage of profiles from the dataset which do not fit well in the assigned cluster. Apart from the 35 to 44 age group, generally the percentage of profiles with negative Si values is below 15%, The age groups with the lowest percent of negative Si values are in the oldest age group. This suggests that older people generally fit better into the three clusters and have less variation in schedules compared to the younger age 73 groups. We also note that, in general, the percentage of profiles with negative Si values are higher on weekdays compared to weekends. On weekdays, nearly 20% of people ages 45 to 54 deviate from the developed cluster profiles, whereas for all other age groups, this is only 4 to 16%. In summary, across the dataset, if not considering the negative-valued Si profiles, the clusters developed represent that the occupancy profiles for approximately 88% of data samples clustered properly in the assigned clusters. As such, this indicates that the three types of schedules represent 88% of people in the United States. The approximately 12% of people with negative Si values represent occupancy schedules which do not fit well within these clusters and may require further analysis to better understand behavior patterns. 74 (a1) (a2) (d1) (d2) -0.6 -0.4 -0.2 0 0.2 0.4 0.6 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 (b1) (b2) (e1) (e2) Elements of each clusters -0.6 -0.4 -0.2 0 0.2 0.4 0.6 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 (f1) (f2) (c1) (c2) -0.6 -0.4 -0.2 0 0.2 0.4 0.6 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 -1 -0.75 -0.5 -0.25 0 0.25 0.5 0.75 1 (g1) (g2) -0.8 -0.4 0 0.4 0.8 -0.8 -0.4 0 0.4 0.8 Silhouette width Figure 3-5. Si (silhouette width) value variation for individuals ages (a) under 25, (b) 25-34, (c) 35-44, (d) 45-54, (e) 55-64, (f) 65-74, (g) over 75 for (1) weekends and (2) weekdays (Note: A positive value of Si width represents a better fit of the data in the clusters; a negative Si value represents data which does not fit well in the assigned cluster 75 25 25 25 Weekday Weekday Weekday Weekend Weekend Weekend Percentage of people 20 20 21 Percentage of people 20 15 15 15 16 15 Percentage of people 15 10 10 13 14 12 12 11 5 5 10 9 10 8 8 0 0 5 Under 25 25 25-34 Under 25-34 35-44 35-44 45-54 45-54 55-64 55-64 65-74 65-74 Over 75 75 Over 4 0 Under 25 25-34 35-44 45-54 55-64 65-74 Over 75 Figure 3-6. Percentage of profiles with negative Si (silhouette width) values by age group 5.4. Individual Occupancy Profile Characteristics As the three types of profiles represent the majority of the U.S. population, all profiles are analyzed in this study. To study the difference between the resulting clusters, the total time duration of absence for each of the clusters was calculated and shown in Figure 7. Each boxplot represents the variation in time duration for the people belonging to each cluster. The median time of absence for the stay-home, day-work, and night-work profiles are 3.0, 11.4 and 10.0 hours respectively on weekends, and 2.4, 10.7 and 10.7 hours respectively on weekdays. Significant variation in the absence time duration can be seen for stay-home and day-work profiles. Both on weekdays and weekends, total time duration of absence for the stay-home profiles are significantly smaller compared to the other two profiles. In addition, for both weekdays and weekends, absence time duration reduces with an increase in age. The mean of absence time duration for the day-work and night-work profiles are similar. However, the distribution of the absence duration is much lower for day-work compared to the nigh- work profile. Similar to the stay-home profile, the day-work profile also has an absence duration that reduces with an increase in age group. 76 Figure 3-7. Variation in the absence time duration for occupants in different age group for (1) stay-home, (2) day-work and (3) night-work profiles on (a) weekdays and (b) weekends The occupancy profiles can be segregated into a combination of different characteristics, assigned to four unique variables in this work. This includes the count of the number of times a person leaves their home (Number of Departures), the time of day when a person departs (Departure Time) and the span of time for which a person is not present at home (Timespan). A fourth variable, Departure Number, represents which departure is being evaluated, e.g. 1st, 2nd, 3rd, 4th, etc… The variation in the Number of Departures, as an example, for people under 25 is shown in Figure 8 for the three developed profile types on weekdays and weekends. The x-axis represents the Number of Departures, and the y-axis represents the percent of people who depart from their home the designated number of times. As seen in Figure 8, a Number of Departures value of 0, 1, 2, 3 and 4 represents the majority of the profiles. For people who follows the stay-home profiles, many of them stay at home for the entire day (i.e. Number of Departures = 0), whereas for the day-work and night-work profiles, people left their home at least once per day. Across both weekday and weekends, the majority of the profiles in the day-work cluster have a Number of Departures of 1, whereas or the night-work cluster this value is 2. This variation in the Number of Departures for all other age groups are shown in the Appendix, which have similar characteristics to those shown in Figure 8. 77 Number of times leaving Number of times leaving Number of times leaving 1 1 100 100 1 100 0.9 0.9 0.9 80 80 0.8 0.8 0.8 80 0.7 0.7 0.7 600.6 60 0.6 60 0.6 0.5 0.5 0.5 400.4 40 0.4 40 0.4 0.3 0.3 0.3 200.2 20 0.2 20 0.2 Percent of people 0.1 0.1 0.1 00 0 0 1 2 3 4 5 6 7 8 9 10 00 0 0 1 2 3 4 5 6 7 8 9 10 00 0 0 1 2 3 4 5 6 7 8 9 10 00 11 22 33 44 55 66 77 88 99 10 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 (a1) (b1) (c1) Number of times leaving Number of times leaving 100 Number of times leaving 1 100 1 100 1 0.9 0.9 0.9 80 0.8 800.8 80 0.8 0.7 0.7 0.7 60 0.6 600.6 60 0.6 0.5 0.5 0.5 40 0.4 400.4 40 0.4 0.3 0.3 0.3 20 0.2 200.2 20 0.2 0.1 0.1 0.1 00 0 00 0 00 0 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 00 11 22 33 44 55 66 77 88 99 10 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 (a2) (b2) (c2) Number of Departures Figure 3-8. Number of times that people under 25 leave their home (Number of Departures) for (a) stay-home, (b) day-work and (c) night-work profiles on (1) weekdays and (2) weekends The variation in the Departure Time and the Timespan that people are gone from their home on weekdays and weekends for the three studied profiles are shown in Figures 9 and 10 respectively. Within each figure, the distributions of Departure Time and the Timespan are shown for the 1st, 2nd, 3rd and 4th departure (Departure Number). For example, for all people who leave 2 or more times from their home, the distribution of Departure Time for the 2nd departure in Figure 9 and 10, under the subheading “2nd departure”. For Departure Time, the horizontal axis shows the time of day of departure, from 12:00 am to 11:59 pm; for Timespan the horizontal axis represents number of hours of absence from home. For the stay-home profiles (Figure 9.a1, 9.b1), during the weekdays the Departure Time varies throughout the day with an increase in the daytime hours. As the Departure Number increases, the Departure Time shifts towards the right, as expected, towards the latter part of the day. The Timespan of absence is short for this type of profile. This makes sense, since people who are staying at home are likely to leave for short trips, such as to go to the store, gym, or other activity rather than the longer durations associated with work hours. Similar distribution of the Departure Number and Departure Time can also be seen on weekends (Figure 10.a1, 10.b1). 78 TimeTime of leave of leave Timespan Timespan of absence of absence 1Once st departure Once 2ndTwo departure times Two times 1st Once departure Once 2nd departure TwoTwo times times 1 100 1 1001 1 1 100 1 100 1 1 0.8 80 0.8 80 0.8 0.8 0.8 80 0.8 80 0.8 0.8 0.6 60 0.6 60 0.6 0.6 0.6 60 0.6 60 0.6 0.6 0.4 40 0.4 40 0.4 0.4 0.4 40 0.4 40 0.4 0.4 0.2 20 0.2 20 0.2 0.2 0.2 20 0.2 20 0.2 0.2 0 000 000 0 0 000 000 0 0 04 48 812 12 16 16 20 20 24 24 0 04 48 812 12 16 16 20 20 24 24 0 0 4 4 8 8 12 12 16 16 20 2024 24 0 0 4 4 8 8 12 1216 1620 2024 24 3rd ThreeThree departure timestimes 4th departure Four Four timestimes Three 3rd Three times times departure 4thFour departure Four times times 1 1001 1001 1 1 100 1 100 1 1 0.8 80 0.8 80 0.8 0.8 80 0.8 0.8 80 0.8 0.8 0.6 60 0.6 60 0.6 0.6 60 0.6 0.6 60 0.6 0.6 0.4 40 0.4 40 0.4 0.4 40 0.4 0.4 40 0.4 0.4 0.2 20 0.2 20 0.2 0.2 20 0.2 0.2 20 0.2 0.2 0 000 000 0 0 000 000 0 0 04 48 812 12 16 16 20 20 24 24 0 04 48 812 12 16 16 20 20 24 24 0 0 4 4 8 8 12 12 16 16 20 2024 24 0 0 4 4 8 8 12 1216 1620 2024 24 TimeTime (a1) of leave of leave Timespan Timespan (b1) of absence of absence 1Once st departure Once 2ndTwo departure times Two times 1stOnce departure Once 2nd Two departure Two timestimes 1 100 1 100 1 1 1 100 1 100 1 1 0.8 80 0.8 80 0.8 0.8 0.8 80 0.8 80 0.8 0.8 0.6 60 0.6 60 0.6 0.6 0.6 60 0.6 60 0.6 0.6 0.4 40 0.4 40 0.4 0.4 0.4 40 0.4 40 0.4 0.4 Percentage distribution 0.2 20 0.2 20 0.2 0.2 0.2 20 0.2 20 0.2 0.2 0 000 000 0 0 000 000 0 0 04 48 812 12 16 16 20 20 24 24 0 04 48 812 12 16 16 20 20 24 24 0 04 48 812 1216 1620 2024 24 0 04 48 812 1216 1620 2024 24 3rd departure ThreeThree timestimes 4th departure Four Four timestimes Three 3rd departure Three times times 4thFour departure Four timestimes 1 100 1 100 1 1 1 100 1 100 1 1 0.8 80 0.8 80 0.8 0.8 80 0.8 0.8 80 0.8 0.8 0.6 60 0.6 60 0.6 0.6 60 0.6 0.6 60 0.6 0.6 0.4 40 0.4 40 0.4 0.4 40 0.4 0.4 40 0.4 0.4 0.2 20 0.2 20 0.2 0.2 20 0.2 0.2 20 0.2 0.2 0 000 00 0 0 0 000 000 0 0 04 48 812 12 16 16 20 20 24 24 0 04 48 812 12 16 16 20 20 24 24 0 04 48 812 1216 1620 2024 24 0 04 48 812 1216 1620 2024 24 Time of leave Time of leave Timespan of absence of (b2) 1 departure st (a2) 2ndTwo departure 1 departure st Timespan absence Once Once times Two times OnceOnce 2Two nd departure Two times times 1 100 100 1 100 1 1 1 1 100 1 1 80 0.8 0.8 80 0.8 0.8 0.8 80 0.8 80 0.8 0.8 60 0.6 0.6 60 0.6 0.6 0.6 60 0.6 60 0.6 0.6 40 0.4 0.4 40 0.4 0.4 0.4 40 0.4 40 0.4 0.4 20 0.2 0.2 20 0.2 0.2 0.2 20 0.2 20 0.2 0.2 0 000 000 0 0 0 000 000 0 04 48 8 12 1216 1620 2024 24 0 04 48 8 12 1216 1620 2024 24 0 04 48 812 12 16 16 20 20 24 24 0 04 48 812 1216 1620 2024 24 3rd Threedeparture times Three times 4th departure FourFour times times 3rd departure 4thFour departure 1 100 1 100 1 1 100 1 ThreeThree timestimes 100 1 Four timestimes 1 1 80 0.8 0.8 80 0.8 0.8 80 0.8 80 0.8 0.8 0.8 60 0.6 0.6 60 0.6 0.6 60 0.6 60 0.6 0.6 0.6 40 0.4 0.4 40 0.4 0.4 40 0.4 40 0.4 0.4 0.4 20 0.2 0.2 20 0.2 0.2 20 0.2 20 0.2 0.2 0.2 0 000 000 0 000 000 0 0 0 04 48 8 12 1216 1620 2024 24 0 04 48 8 12 1216 1620 2024 24 0 04 48 812 12 16 16 20 20 24 24 0 04 48 812 1216 1620 2024 24 (a3) Time of day (b3) (b3) Figure 3-9. (a) Departure Time and (b) Timespan that people are gone from their home for (1) stay-home, (2) day-work and (3) night-work profiles for people under 25 on weekdays 79 TimeTime of leave of leave Timespan Timespan of absence of absence 1Once st departure Once 2ndTwo departure Two timestimes 1stOnce departure Once 2nd departure TwoTwo times times 1 1001 1001 1 1 100 1 100 1 1 0.8 80 0.8 80 0.8 0.8 0.8 80 0.8 80 0.8 0.8 0.6 60 0.6 60 0.6 0.6 0.6 60 0.6 60 0.6 0.6 0.4 40 0.4 40 0.4 0.4 0.4 40 0.4 40 0.4 0.4 0.2 20 0.2 20 0.2 0.2 0.2 20 0.2 20 0.2 0.2 0 000 000 0 0 000 000 0 0 04 48 812 1216 1620 2024 24 0 04 48 812 1216 1620 2024 24 0 0 4 4 8 8 12 1216 1620 2024 24 0 0 4 4 8 8 12 1216 1620 2024 24 3rd Three departure Three timestimes 4th departure FourFour timestimes Three 3rd departure Three times times 4th departure Four Four times times 1 1001 1001 1 1 1001 100 1 1 0.8 80 0.8 80 0.8 0.8 0.8 80 0.8 80 0.8 0.8 0.6 60 0.6 60 0.6 0.6 0.6 60 0.6 60 0.6 0.6 0.4 40 0.4 40 0.4 0.4 0.4 40 0.4 40 0.4 0.4 0.2 20 0.2 20 0.2 0.2 0.2 20 0.2 20 0.2 0.2 0 000 000 0 0 000 000 0 0 04 48 812 1216 1620 2024 24 0 04 48 812 1216 1620 2024 24 0 0 4 4 8 8 12 1216 1620 2024 24 0 0 4 4 8 8 12 1216 1620 2024 24 TimeTime of leave of leave Timespan Timespan of absence of absence 1st departure (a1) 2nd departure (b1) 100 Once Once Two Two times times 1stdeparture OnceOnce 2nd departure Two times Two times 1 1 100 100 1 100 1 1 1 100 1 1 80 0.8 0.8 80 80 0.8 80 0.8 0.8 0.8 80 0.8 0.8 0.6 60 0.6 60 0.6 0.6 60 60 0.6 0.6 60 0.6 0.6 40 0.4 0.4 40 40 0.4 40 0.4 40 Percentage distribution 0.4 0.4 0.4 0.4 20 0.2 0.2 0.2 20 0.2 0.2 20 20 0.2 20 0.2 0.2 000 000 0 00 000 0 04 48 8 12 16 20 24 04 48 8 12 16 20 0 24 0 000 0 0 12 16 20 24 0 12 16 20 24 04 48 812 12 16 16 20 20 24 24 0 04 48 812 12 16 16 20 20 24 24 3rdtimes departure ThreeThree times 4th departure Four times Four times 33rdrddeparture departure 4th departure 100 1 1 ThreeThree timestimes Four Four timestimes 1 100 1 1 100 1 1001 1 0.8 80 0.8 80 0.8 0.8 0.8 80 0.8 80 0.8 0.8 0.6 60 0.6 60 0.6 0.6 0.6 60 0.6 60 0.6 0.6 0.4 40 0.4 40 0.4 0.4 0.4 40 0.4 40 0.4 0.4 0.2 20 0.2 20 0.2 0.2 0.2 20 0.2 20 0.2 0.2 0 000 000 0 000 000 0 0 0 04 48 8 12 12 16 16 20 20 24 24 0 04 48 8 12 12 16 16 20 20 24 24 0 04 48 812 12 16 16 20 20 24 24 0 04 48 812 12 16 16 20 20 24 24 Time(a2) of leave Time of leave Timespan Timespan of(b2) absence of absence 1st departure 2nd departure 1Once st departure Once 2ndTwo departure times Two times Once 100 Once Two times 100 Two times 1001 1 1 1 100 11 1001 1 80 0.8 80 80 0.8 80 0.8 0.8 80 0.8 0.8 0.8 0.8 60 0.6 60 60 0.6 60 0.6 0.6 60 0.6 0.6 0.6 0.6 40 0.4 40 40 0.4 40 0.4 0.4 40 0.4 0.4 0.4 0.4 20 0.2 20 20 0.2 20 0.2 0.2 20 0.2 0.2 0.2 0.2 000 000 0 0 00 0000 000 0 0 4 8 12 16 20 24 0 04 48 812 1216 1620 2024 24 0 004 448 8812 12 16 12 16 20 16 20 24 20 24 24 0 04 48 812 12 16 16 20 20 24 24 Three3 rd departure 4Four th departure 3rd departure Three times 4th departure Three Four times times Three timestimes Four times Four timestimes 1001 100 1 1 1 100 11 1001 1 80 0.8 80 0.8 0.8 0.8 80 0.8 0.8 0.8 80 0.8 60 0.6 60 0.6 0.6 0.6 60 0.6 0.6 0.6 60 0.6 40 0.4 40 0.4 0.4 0.4 40 0.4 0.4 0.4 40 0.4 20 0.2 20 0.2 0.2 0.2 20 0.2 0.2 20 0.2 0.2 000 000 0 0 0000 000 0 0 4 8 12 16 20 24 0 04 48 812 1216 1620 2024 24 0 004 448 8812 12 16 12 16 20 16 20 24 20 24 24 0 04 48 812 12 16 16 20 20 24 24 (a3) Time of day (b3) Figure 3-10. (a) Departure Time and (b) Timespan distribution for (1) stay-home, (2) day-work and (3) night-work profiles for people under 25 on weekends 80 For the day-work profiles, Departure Time (Figure 9.a2 and 10.a2) peaks in the early morning for a Departure Number of 1. For this profile, on weekdays especially, and to a lesser extent on weekends, Timespan values (Figure 9.b2 and 10.b2) are clustered around an absence of 8 to 12 hours. Different characteristics can be seen for the night-work profiles for Departure Time (Figure 9.a3 and 10.a3) and Timespan (Figure 9.b3 and 10.b3). For the 1st departure (Departure Number = 1), the Departure Time varies nearly uniformly whereas for the 2nd departure, it generally occurs during the latter part of the day. For this profile, the Timespan of absence is, on average, approximately 4 hours for the 1st departure, whereas for the 2nd departure, there is a much broader range of Timespan values across the data. Similar results to the weekday profile are also seen on weekends, although the Timespan values are generally smaller on weekends. Similar characteristics are also seen for people belonging to all other age groups for both weekdays and weekends (see Appendix). These different characteristics of the Departure Number, Departure Time and Timespan can be used together to calculate the occupancy schedules for residential buildings in the United States. The variation in the absence time duration provides a gauge of the relative difference in energy use across the profile types due to occupant-based end use loads, the most common of which is HVAC units which can be dependent on the presence and absence of occupancy [10]. 6. Conclusions In this study, ATUS data is studied to analyze the occupancy profiles for individuals in the United States. Cluster analysis was completed initially to determine different patterns in occupancy schedules for people in different age group and whether its weekday or weekends. After that, the accuracy of the cluster profiles was evaluated and the variations in the profiles were studied. Different characteristics of the schedules were also analyzed. The overall key findings of this study can be summarized as follows: • Occupancy schedules of individuals living in the U.S. can be represented using three major profile types including, Day-work, Stay-home and Night-work. These three profiles are found across all individual subgroups studied, irrespective of age group and weekday and weekend considerations. 81 • Though the patterns are similar by age group and weekdays/weekends, the percentage of people who fall into each of these schedule categories varies significantly based on the age group of the occupants and weekday and weekends. • On weekends, the majority of people follow a Stay-home schedule, irrespective of their age. • On weekdays, the percentage of people who follow the Day-work profile increases with age, peaking at the age group of 35-44; it then begins to decrease with increasing age. • People in the age groups of 65 to 74 and over 75 follows very similar occupancy profiles with similar characteristics. • The three developed profiles represent a strong fit for 88% of the U.S. population. The remaining 12% do not fit as well within the developed clusters and require further study to better understand their occupancy patterns and behaviors. • Profile characteristics are similar for the Stay-home profile on both weekdays and weekends. However, the Day-work profile experiences higher variations in profile when comparing weekdays and weekends. • The occupancy profiles can be characterized using four main parameters: Number of Departures, Departure Time, Timespan and Departure Number. • All four parameters vary for different profile types. Variations in these three parameters can also be seen for different age groups and for weekdays and weekends. The result of this study can be used to predict the occupancy schedule for a specific person based on their age and general information of the type of profiles the person follows. The total absence time duration distribution for each of the clusters demonstrates the significant variation in occupancy schedules among different profiles. This can influence the overall energy performance of the building and its energy consuming systems, particularly when occupant-based controls are used to adjust a home’s energy consuming systems. One limitation of this study is that ATUS data was utilized which is self-reported data; human error can influence self-reported data. In addition, other occupant and/or household characteristics such as financial status, can also help to further study to represent the occupant profiles in more detail. Several recent papers have discussed the comparison between other nationwide datasets such as the Residential Energy Consumption Survey data and ATUS data [40, 55]. Thus, to further validate these findings related to occupancy 82 schedules, it would be beneficial to compare the results with the Residential Energy Consumption Survey data or other national level datasets. However, due to unavailability of occupancy scheduling data in this and other datasets, the comparison cannot be completed at this time. For future studies, the result of this work can be used to create an improved, stochastic occupancy simulator for residential buildings in the United States that varies by age group. The impact of the different occupancy schedules on the energy performance of a residential building could also be analyzed using this occupancy data as input. 7. Acknowledgements The information, data, or work presented herein was funded in part by the Advanced Research Projects Agency‐Energy (ARPA‐E), U.S. Department of Energy, under Award Number DE‐ AR0001256. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof. Additionally, the authors would like to thank Nicholas Steinmetz for his professional and constructive help on statistical methods and techniques. 83 REFERENCES [1] Saha H., Florita, A. R., Henze, G. P., Sarkar S., “Occupancy sensing in buildings: A review of data analytics approaches”. Energy&Buildings188–189(2019)278–285. [2] Yoshino H., Hong T., Nord N., IEA EBC annex 53: total energy use in buildings—analysis and evaluation methods, Energy Build 152 (2017) 124–136, doi: 10. 1016/j.enbuild.2017.07.038 . [3] Yan D., Brien W.O., Hong T., Feng X., Gunay H.B., Tahmasebi F., “A. Mahdavi, Occupant behavior modeling for building performance simulation: Current state and future challenges”, 107 (2015) 264–278. doi:10.1016/j.enbuild.2015.08.032. [4] Yu Z., Fung B.C.M., Haghighat F., Yoshino H., Morofsky E., “A systematic procedure to study the influence of occupant behavior on building energy consumption”. Energy Build, 43 (2011), pp. 1409-1417, 10.1016/j.enbuild.2011.02.002. [5] Lo L.J., Novoselac A., “Localized air-conditioning with occupancy control in an open office”. Energy Build., 42 (7) (2010), pp. 1120-1128. [6] Yun G.Y., Steemers K., Behavioural, physical and socio-economic factors in household cooling energy consumption, Appl. Energy. 88 (2011) 2191–2200, doi: 10.1016/j.apenergy.2011.01.010 [7] Iwashita, G, and Hiroshi Akasaka. "The effects of human behavior on natural ventilation rate and indoor air environment in summer—a field study in southern Japan." Energy and Buildings 25, no. 3 (1997): 195-205. [8] Li, Z.-J., Jiang, Y., Wei, Q.-P., 2014, “Survey on energy consumption of air conditioning in summer in a residential building in Beijing”, Heat. Vent. Air Cond., 37 (2007), pp. 46-51. [9] An J, Yan D, Hong T. Clustering and statistical analyses of air-conditioning intensity and use patterns in residential buildings[J]. Energy and Buildings, 2018, 174: 214-227. [10] Xia D, Lou S, Huang Y, et al. A study on occupant behaviour related to air-conditioning usage in residential buildings[J]. Energy and Buildings, 2019, 203: 109446. [11] Santin, O. Guerra, Itard, L., Visscher, H., “The effect of occupancy and building characteristics on energy use for space and water heating in Dutch residential stock”, Energy Build, 41 (2009), pp. 1223-1232, 10.1016/j.enbuild.2009.07.002. [12] Wilke, U., “Probabilistic Bottom-up Modelling of Occupancy and Activities to Predict Electricity Demand in Residential Buildings”. PhD Thesis. 5673 172 (2013), 10.5075/epfl-thesis- 5673. [13] Popoola, O. M. 2018. “Computational Intelligence Modelling Based on Variables Interlinked with Behavioral Tendencies for Energy Usage Profile – A Necessity.” Renewable and Sustainable Energy Reviews 82 (May 2017): 60–72. https://doi.org/10.1016/j.rser.2017.09.020. 84 [14] Happle, G, Jimeno A. F., and Arno S. "A review on occupant behavior in urban building energy models." Energy and Buildings 174 (2018): 276-292. [15] Chen, Z., Chaoyang J., and Lihua X.. "Building occupancy estimation and detection: A review." Energy and Buildings 169 (2018): 260-270. [16] Page, J., Robinson, D., Morel, N., Scartezzini, J.-L. “A generalised stochastic model for the simulation of occupant presence”. Energy and Buildings 40 (2008) 83–98. [17] Richardson, I., Thomson M., Infield, D., “A high-resolution domestic building occupancy model for energy demand simulations”. Energy and Buildings 40 (2008) 1560–1566. [18] Adamopoulou, A. A., Tryferidis, A. M., Tzovaras, D. K., “A context-aware method for building occupancy prediction.” Energy and Buildings 110 (2016) 229–244. [19] Chen, Z., Jinming X., and Yeng C. S., "Modeling regular occupancy in commercial buildings using stochastic models." Energy and Buildings 103 (2015): 216-223. [20] Schwartz, Z., Uysal, M., Webb, T., Altin, M., “Hotel daily occupancy forecasting with competitive sets:a recursive algorithm”.Int. J. Contemp. Hosp. Manag., 28 (2) (2016), pp.267-285. [21] Chen D., Barker, S., Subbaswamy, A., Irwin, D., Shenoy, P, "Non-intrusive occupancy monitoring using smart meters." In Proceedings of the 5th ACM Workshop on Embedded Systems For Energy-Efficient Buildings, pp.1-8. 2013 [22] Buttitta, G., William JN T., and Donal F., "Clustering of household occupancy profiles for archetype building models." Energy Procedia (2016). [23] Zhao, J., Bertrand L., Lam, K. P., Yun, R., and Loftness, V., "Occupant behavior and schedule modeling for building energy simulation through office appliance power consumption data mining." Energy and Buildings 82 (2014): 341-355. [24] Lam, K. P., Höynck, M., Dong, B., Andrews, B., Chiou, Y., Zhang, R., Benitez, D., and Choi, J., "Occupancy detection through an extensive environmental sensor network in an open-plan office building." IBPSA Building Simulation 145 (2009): 1452-1459. [25] Amato, G., Carrara, F., Falchi, F., Gennaro, C., Meghini, C., Vairo, C., "Deep learning for decentralized parking lot occupancy detection." Expert Systems with Applications 72 (2017): 327- 334. [26] Chen, Z., Zhao, R., Zhu, Q., Masood, M. K., Soh, Y. C., and Mao, K., "Building occupancy estimation with environmental sensors via CDBLSTM." IEEE Transactions on Industrial Electronics 64, no. 12 (2017): 9549-9559. [27] Balvedi, B. F., Ghisi, E., and Lamberts, R., "A review of occupant behaviour in residential buildings." Energy and Buildings 174 (2018): 495-505. 85 [28] Carpino, C., Fajilla, G., Gaudio, A., Mora, D., and Simone, M. D. "Application of survey on energy consumption and occupancy in residential buildings. An experience in Southern Italy." Energy Procedia 148 (2018): 1082-1089. [29] Chen, S., Yang, W., Yoshino, H., Levine, M. D., Newhouse, K., and Hinge, A., "Definition of occupant behavior in residential buildings and its application to behavior analysis in case studies." Energy and Buildings104(2015):1-13. [30] Hu, S., Yan, D., An, J., Guo, S., Qian, M., “Investigation and analysis of Chinese residential building occupancy with large-scale questionnaire surveys”. Energy & Buildings 193 (2019) 289– 304. [31] C. National Bureau of Statistics 2008 Time Use Survey in China, China Statistics Press (2009), https://unstats.un.org/unsd/demographic/sconcerns/tuse/Country/china/Time Use Survey in China.pdf [32] Statistics Bureau, Japan, https://www.stat.go.jp/english/data/shakai/index.html [33] Aerts, D., J. Minnen, I. Glorieux, I. Wouters, and F. Descamps. "A method for the identification and modelling of realistic domestic occupancy sequences for building energy demand simulations and peer comparison." Building and environment 75 (2014): 67-78. [34] Blight, T. S., and David A. C., "Sensitivity analysis of the effect of occupant behaviour on the energy consumption of passive house dwellings." Energy and Buildings 66 (2013): 183-192. [35] McKenna, E., Krawczynski, M., and Thomson, M., "Four-state domestic building occupancy model for energy demand simulations." Energy and Buildings 96 (2015): 30-39. [36] López-Rodríguez, M. A., Santiago, I., Trillo-Montero, D., Torriti, J., Moreno-Munoz, A., "Analysis and modeling of active occupancy of the residential sector in Spain: An indicator of residential electricity consumption." Energy Policy 62 (2013): 742-751. [37] Chiou, Y.-S., Carley, K. M., Davidson, C. I., and Johnson, M. P., "A high spatial resolution residential energy model based on American Time Use Survey data and the bootstrap sampling method." Energy and Buildings 43, no. 12 (2011): 3528-3538. [38] U.S. Bureau of Labor Statistics, American Time Use Survey, (2009). http:// www.bls.gov/tus/#data (Accessed on 10th June, 2018.). [39] Gershuny, J., Sullivan, O. (2017). United Kingdom Time Use Survey, 2014-2015. Centre for Time Use Research, University of Oxford. [data collection]. UK Data Service. SN: 8128, http://doi.org/10.5255/UKDA-SN-8128-1. [40] Mitra, D., Steinmetz, N., Chu, Y., and Cetin, K. S., "Typical occupancy profiles and behaviors in residential buildings in the United States." Energy and Buildings 210 (2020): 109713. [41] U.S. Residential Energy Consumption Survey. (2015). https://www.eia.gov/consumption/residential/index.php (Accessed on 10th June, 2018.). 86 [42] Liao, W. T, Liao TW. “Clustering of time series data – a survey”. Pattern Recognit 2005;38(11):1857–74., http://dx.doi.org/10.1016/ j.patcog.2005.01.025. [43] Reddy TA. “Chapter 8: Classification and clustering methods”. In: Applied data analysis and modeling for energy engineers and scientists; 2011. p. 231–251. [44] Diao, L, Sun Y, Chen Z, and Chen J. 2017. “Modeling Energy Consumption in Residential Buildings: A Bottom-up Analysis Based on Occupant Behavior Pattern Clustering and Stochastic Simulation.” Energy and Buildings 147: 47–66. https://doi.org/10.1016/j.enbuild.2017.04.072. [45] Chicco G. “Overview and performance assessment of the clustering methods for electrical load pattern grouping”. Energy 2012;42(1):68–80. http://dx.doi.org/ 10.1016/j.energy.2011.12.031. [46] Kim YI, Ko JM, Choi SH. “Methods for generating TLPs (typical load profiles) for smart grid-based energy programs”. In: 2011 IEEE symposium on computational intelligence applications in smart grid (CIASG). IEEE; 2011. p. 1–6. ISBN 978-1-4244-9893-2. http://dx.doi.org/10.1109/ CIASG.2011.5953331. [47] Kwac J., Flora J., Rajagopal R., “Household energy consumption segmentation using hourly data”. IEEE Trans Smart Grid 2014;5(1):420–30. http://dx.doi.org/ 10.1109/TSG.2013.2278477. [48] Likas, A., Vlassis, N., Verbeek, J. J., “The global k-means clustering algorithm.” Pattern Recognition 36 (2003) 451–461. [49] Charrad, M., Ghazzali, N., Boiteau, V., Niknafs, A., and Charrad, M. M., "Package ‘nbclust’." Journal of statistical software 61 (2014): 1-36. [50] Brock, G., Pihur, V., Datta, S., Datta, S., "clValid, an R package for cluster validation." Journal of Statistical Software (Brock et al., March 2008) (2011). [51] Team, R. Core. "R: A language and environment for statistical computing." (2013): 201. [52] Kaufman, L., Rousseeuw, P. J., Finding Groups in Data: An Introduction to Cluster Analysis, Wiley series in probability and mathematical statistics.,1990, John Wiley and Sons Inc. [53] Reynolds, A. P., Richards, G., Beatriz de la Iglesia, Victor J. Rayward-Smith. "Clustering rules: a comparison of partitioning and hierarchical clustering algorithms." Journal of Mathematical Modelling and Algorithms 5, no. 4 (2006): 475-504. [54] Deru, M., Field K., Studer D., Benne K., Griffith B., Torcellini P., Liu B., et al. "US Department of Energy commercial reference building models of the national building stock." (2011): 1. [55] Mitra, D., Chu, Y., Steinmetz, N., Cetin, K., Lovejoy, J., Kremer, P. (2019) Defining Typical Occupancy Schedules and Behaviors in Residential Buildings Using the American Time Use Survey, 2019 ASHRAE Annual Conference, Kansas City, MO, June 22-26, 2019 87 APPENDIX Table A2. Weekday profiles for Stay-home Time Under 25 25-34 35-44 45-54 55-64 65-74 Over 75 Time Under 25 25-34 35-44 45-54 55-64 65-74 Over 75 0:00 0.97 0.99 0.99 0.98 0.99 1.00 1.00 6:00 0.99 0.88 0.96 0.95 0.96 0.97 0.98 0:05 0.98 0.99 0.99 0.98 0.99 1.00 1.00 6:05 0.99 0.88 0.96 0.95 0.96 0.97 0.98 0:10 0.98 0.99 0.99 0.99 0.99 1.00 1.00 6:10 0.98 0.88 0.96 0.95 0.96 0.97 0.98 0:15 0.98 0.99 0.99 0.99 0.99 1.00 1.00 6:15 0.98 0.88 0.96 0.95 0.95 0.97 0.98 0:20 0.99 0.99 0.99 0.99 0.99 1.00 1.00 6:20 0.98 0.88 0.96 0.95 0.95 0.97 0.97 0:25 0.99 0.99 1.00 0.99 0.99 1.00 1.00 6:25 0.98 0.88 0.96 0.95 0.95 0.97 0.97 0:30 0.99 0.99 1.00 0.99 0.99 1.00 1.00 6:30 0.97 0.87 0.95 0.94 0.93 0.96 0.96 0:35 0.99 0.99 1.00 0.99 0.99 1.00 1.00 6:35 0.97 0.87 0.95 0.94 0.93 0.96 0.96 0:40 0.99 0.99 1.00 0.99 0.99 1.00 1.00 6:40 0.97 0.87 0.95 0.94 0.92 0.96 0.96 0:45 0.99 0.99 1.00 0.99 0.99 1.00 1.00 6:45 0.97 0.86 0.94 0.93 0.91 0.95 0.96 0:50 0.99 1.00 1.00 0.99 0.99 1.00 1.00 6:50 0.96 0.86 0.94 0.93 0.91 0.95 0.96 0:55 0.99 1.00 1.00 0.99 0.99 1.00 1.00 6:55 0.96 0.86 0.94 0.93 0.91 0.95 0.96 1:00 0.99 0.99 1.00 0.99 1.00 1.00 1.00 7:00 0.95 0.83 0.92 0.91 0.87 0.93 0.95 1:05 0.99 0.99 1.00 0.99 1.00 1.00 1.00 7:05 0.94 0.83 0.91 0.91 0.87 0.93 0.94 1:10 0.99 0.99 1.00 0.99 1.00 1.00 1.00 7:10 0.94 0.83 0.91 0.91 0.87 0.93 0.94 1:15 0.99 0.99 1.00 0.99 1.00 1.00 1.00 7:15 0.93 0.82 0.90 0.90 0.85 0.92 0.94 1:20 0.99 1.00 1.00 0.99 1.00 1.00 1.00 7:20 0.93 0.82 0.90 0.90 0.85 0.92 0.94 1:25 0.99 1.00 1.00 0.99 1.00 1.00 1.00 7:25 0.93 0.82 0.90 0.90 0.85 0.92 0.94 1:30 0.99 1.00 1.00 1.00 1.00 1.00 1.00 7:30 0.91 0.79 0.87 0.88 0.82 0.91 0.93 1:35 0.99 1.00 1.00 1.00 1.00 1.00 1.00 7:35 0.91 0.79 0.87 0.88 0.82 0.91 0.92 1:40 0.99 1.00 1.00 1.00 1.00 1.00 1.00 7:40 0.91 0.79 0.87 0.88 0.81 0.90 0.92 1:45 0.99 1.00 1.00 1.00 1.00 1.00 1.00 7:45 0.90 0.78 0.87 0.87 0.80 0.90 0.92 1:50 1.00 1.00 1.00 1.00 1.00 1.00 1.00 7:50 0.90 0.78 0.87 0.87 0.80 0.89 0.92 1:55 1.00 1.00 1.00 1.00 1.00 1.00 1.00 7:55 0.90 0.78 0.87 0.87 0.80 0.89 0.91 2:00 0.99 0.99 1.00 0.99 1.00 1.00 1.00 8:00 0.88 0.75 0.84 0.85 0.77 0.87 0.90 2:05 0.99 0.99 1.00 1.00 1.00 1.00 1.00 8:05 0.88 0.76 0.85 0.86 0.76 0.87 0.89 2:10 0.99 0.99 1.00 1.00 1.00 1.00 1.00 8:10 0.88 0.76 0.85 0.86 0.76 0.86 0.89 2:15 0.99 0.99 1.00 1.00 1.00 1.00 1.00 8:15 0.87 0.76 0.85 0.86 0.75 0.86 0.89 2:20 1.00 0.99 1.00 1.00 1.00 1.00 1.00 8:20 0.87 0.75 0.85 0.86 0.75 0.85 0.89 2:25 1.00 0.99 1.00 1.00 1.00 1.00 1.00 8:25 0.87 0.75 0.84 0.86 0.75 0.85 0.88 2:30 1.00 0.99 1.00 1.00 1.00 1.00 1.00 8:30 0.86 0.74 0.83 0.85 0.73 0.83 0.86 2:35 1.00 0.99 1.00 1.00 1.00 1.00 1.00 8:35 0.85 0.74 0.83 0.85 0.73 0.83 0.86 2:40 1.00 0.99 1.00 1.00 1.00 1.00 1.00 8:40 0.85 0.74 0.83 0.85 0.73 0.83 0.86 2:45 1.00 0.99 1.00 1.00 1.00 1.00 1.00 8:45 0.85 0.73 0.83 0.85 0.73 0.82 0.86 2:50 1.00 0.99 1.00 1.00 1.00 1.00 1.00 8:50 0.84 0.74 0.83 0.86 0.72 0.82 0.85 2:55 1.00 0.99 1.00 1.00 1.00 1.00 1.00 8:55 0.84 0.74 0.83 0.86 0.72 0.82 0.85 3:00 1.00 0.99 1.00 0.99 1.00 1.00 1.00 9:00 0.81 0.72 0.80 0.84 0.70 0.78 0.82 3:05 1.00 0.99 1.00 1.00 1.00 1.00 1.00 9:05 0.81 0.72 0.81 0.85 0.70 0.78 0.82 3:10 1.00 0.99 1.00 0.99 1.00 1.00 1.00 9:10 0.81 0.72 0.81 0.85 0.70 0.78 0.82 3:15 1.00 0.99 1.00 0.99 1.00 1.00 1.00 9:15 0.80 0.72 0.81 0.85 0.69 0.77 0.81 3:20 1.00 0.99 1.00 0.99 1.00 1.00 1.00 9:20 0.80 0.71 0.81 0.86 0.69 0.77 0.81 3:25 1.00 0.99 1.00 0.99 1.00 1.00 1.00 9:25 0.80 0.71 0.81 0.86 0.69 0.77 0.81 3:30 1.00 0.99 0.99 0.99 1.00 1.00 1.00 9:30 0.77 0.70 0.80 0.86 0.68 0.75 0.79 3:35 1.00 0.99 0.99 0.99 1.00 1.00 1.00 9:35 0.77 0.69 0.79 0.86 0.68 0.75 0.80 3:40 1.00 0.99 0.99 0.99 1.00 1.00 1.00 9:40 0.77 0.70 0.79 0.86 0.68 0.75 0.80 3:45 1.00 0.99 0.99 0.99 0.99 1.00 1.00 9:45 0.76 0.69 0.79 0.86 0.67 0.75 0.79 3:50 1.00 0.99 0.99 0.99 0.99 1.00 1.00 9:50 0.76 0.69 0.79 0.87 0.67 0.75 0.79 3:55 1.00 0.99 0.99 0.99 0.99 1.00 1.00 9:55 0.76 0.69 0.79 0.87 0.67 0.75 0.79 4:00 0.99 0.96 0.97 0.97 0.99 1.00 1.00 10:00 0.73 0.66 0.77 0.87 0.65 0.72 0.77 4:05 0.99 0.96 0.97 0.97 0.99 1.00 1.00 10:05 0.73 0.66 0.77 0.87 0.65 0.73 0.77 4:10 0.99 0.96 0.97 0.97 0.99 0.99 1.00 10:10 0.73 0.67 0.77 0.88 0.65 0.73 0.77 4:15 0.99 0.96 0.97 0.97 0.99 1.00 1.00 10:15 0.73 0.66 0.76 0.88 0.65 0.72 0.77 4:20 1.00 0.96 0.97 0.97 0.99 1.00 1.00 10:20 0.72 0.66 0.76 0.89 0.65 0.73 0.77 4:25 1.00 0.96 0.97 0.97 0.99 1.00 1.00 10:25 0.72 0.66 0.76 0.89 0.65 0.73 0.78 4:30 1.00 0.95 0.97 0.96 0.99 0.99 1.00 10:30 0.70 0.65 0.75 0.88 0.64 0.71 0.76 4:35 1.00 0.95 0.97 0.96 0.99 0.99 1.00 10:35 0.70 0.65 0.75 0.89 0.64 0.71 0.77 4:40 1.00 0.95 0.97 0.96 0.99 0.99 1.00 10:40 0.71 0.65 0.75 0.89 0.64 0.72 0.77 4:45 1.00 0.94 0.97 0.96 0.99 0.99 0.99 10:45 0.70 0.65 0.75 0.89 0.63 0.71 0.77 4:50 1.00 0.94 0.97 0.96 0.99 0.99 0.99 10:50 0.70 0.65 0.75 0.89 0.64 0.71 0.77 4:55 1.00 0.94 0.97 0.96 0.99 0.99 0.99 10:55 0.70 0.65 0.75 0.89 0.64 0.72 0.77 5:00 1.00 0.93 0.97 0.96 0.98 0.99 0.99 11:00 0.67 0.63 0.73 0.88 0.62 0.69 0.76 5:05 1.00 0.92 0.97 0.96 0.98 0.99 0.99 11:05 0.67 0.63 0.73 0.88 0.63 0.70 0.76 5:10 1.00 0.92 0.97 0.96 0.98 0.99 0.99 11:10 0.67 0.63 0.74 0.88 0.63 0.70 0.77 5:15 1.00 0.92 0.97 0.96 0.98 0.99 0.99 11:15 0.67 0.63 0.74 0.88 0.63 0.70 0.77 5:20 1.00 0.92 0.97 0.96 0.98 0.99 0.99 11:20 0.67 0.64 0.74 0.89 0.63 0.70 0.77 5:25 1.00 0.92 0.97 0.96 0.98 0.99 0.99 11:25 0.67 0.64 0.74 0.89 0.63 0.71 0.77 5:30 0.99 0.90 0.96 0.96 0.98 0.99 0.99 11:30 0.65 0.62 0.73 0.88 0.62 0.70 0.77 5:35 0.99 0.90 0.97 0.96 0.98 0.99 0.99 11:35 0.66 0.62 0.73 0.89 0.63 0.70 0.77 5:40 1.00 0.90 0.97 0.96 0.98 0.99 0.99 11:40 0.66 0.63 0.74 0.89 0.63 0.70 0.78 5:45 0.99 0.89 0.96 0.96 0.97 0.98 0.99 11:45 0.66 0.63 0.73 0.88 0.63 0.70 0.78 5:50 0.99 0.89 0.96 0.96 0.97 0.98 0.99 11:50 0.66 0.63 0.74 0.88 0.63 0.71 0.78 5:55 0.99 0.89 0.96 0.96 0.97 0.98 0.99 11:55 0.66 0.63 0.74 0.89 0.63 0.71 0.79 88 Table A2(cont’d) Time Under 25 25-34 35-44 45-54 55-64 65-74 Over 75 Time Under 25 25-34 35-44 45-54 55-64 65-74 Over 75 12:00 0.63 0.61 0.72 0.86 0.61 0.69 0.77 18:00 0.65 0.77 0.85 0.71 0.83 0.83 0.86 12:05 0.64 0.62 0.72 0.87 0.61 0.69 0.77 18:05 0.65 0.78 0.86 0.71 0.83 0.84 0.86 12:10 0.64 0.62 0.73 0.87 0.62 0.70 0.78 18:10 0.65 0.79 0.87 0.72 0.83 0.84 0.86 12:15 0.64 0.63 0.73 0.87 0.63 0.70 0.78 18:15 0.65 0.79 0.87 0.71 0.84 0.84 0.86 12:20 0.65 0.64 0.74 0.87 0.63 0.70 0.79 18:20 0.65 0.79 0.87 0.72 0.84 0.85 0.86 12:25 0.65 0.64 0.74 0.87 0.63 0.70 0.79 18:25 0.66 0.80 0.87 0.72 0.84 0.85 0.87 12:30 0.64 0.63 0.73 0.86 0.63 0.69 0.78 18:30 0.66 0.80 0.86 0.72 0.84 0.85 0.86 12:35 0.65 0.63 0.74 0.86 0.63 0.70 0.78 18:35 0.66 0.80 0.86 0.72 0.84 0.85 0.86 12:40 0.65 0.64 0.75 0.87 0.63 0.70 0.78 18:40 0.66 0.80 0.86 0.72 0.85 0.85 0.86 12:45 0.65 0.64 0.75 0.87 0.63 0.70 0.79 18:45 0.66 0.81 0.86 0.72 0.85 0.85 0.86 12:50 0.65 0.64 0.76 0.87 0.63 0.70 0.79 18:50 0.66 0.81 0.87 0.72 0.85 0.86 0.86 12:55 0.65 0.64 0.76 0.87 0.64 0.71 0.79 18:55 0.67 0.81 0.87 0.72 0.85 0.86 0.86 13:00 0.63 0.63 0.75 0.85 0.63 0.69 0.78 19:00 0.65 0.81 0.86 0.71 0.84 0.86 0.86 13:05 0.64 0.64 0.76 0.86 0.64 0.70 0.78 19:05 0.66 0.82 0.86 0.71 0.85 0.87 0.86 13:10 0.64 0.64 0.77 0.86 0.64 0.70 0.79 19:10 0.66 0.82 0.87 0.72 0.85 0.87 0.87 13:15 0.64 0.65 0.77 0.86 0.64 0.70 0.79 19:15 0.67 0.84 0.87 0.73 0.85 0.88 0.87 13:20 0.65 0.65 0.77 0.86 0.64 0.71 0.79 19:20 0.67 0.84 0.87 0.73 0.85 0.88 0.87 13:25 0.65 0.65 0.78 0.86 0.65 0.71 0.80 19:25 0.68 0.84 0.88 0.73 0.86 0.88 0.87 13:30 0.64 0.65 0.77 0.85 0.64 0.70 0.79 19:30 0.68 0.84 0.87 0.73 0.86 0.88 0.87 13:35 0.64 0.65 0.78 0.85 0.64 0.71 0.80 19:35 0.68 0.85 0.88 0.74 0.86 0.89 0.88 13:40 0.64 0.65 0.78 0.85 0.65 0.71 0.80 19:40 0.69 0.85 0.88 0.74 0.87 0.89 0.88 13:45 0.65 0.65 0.78 0.85 0.65 0.72 0.81 19:45 0.69 0.85 0.88 0.74 0.87 0.90 0.88 13:50 0.65 0.66 0.78 0.85 0.66 0.72 0.81 19:50 0.70 0.86 0.88 0.75 0.88 0.90 0.88 13:55 0.65 0.66 0.78 0.85 0.66 0.73 0.82 19:55 0.70 0.87 0.88 0.75 0.88 0.91 0.89 14:00 0.63 0.65 0.77 0.82 0.64 0.71 0.80 20:00 0.70 0.87 0.88 0.76 0.88 0.91 0.89 14:05 0.63 0.65 0.77 0.82 0.65 0.71 0.81 20:05 0.71 0.87 0.89 0.76 0.88 0.91 0.89 14:10 0.64 0.66 0.77 0.83 0.65 0.72 0.81 20:10 0.71 0.88 0.89 0.77 0.89 0.92 0.90 14:15 0.64 0.66 0.78 0.83 0.65 0.72 0.82 20:15 0.72 0.88 0.89 0.78 0.89 0.92 0.90 14:20 0.64 0.67 0.78 0.83 0.66 0.73 0.82 20:20 0.73 0.89 0.89 0.78 0.89 0.93 0.90 14:25 0.64 0.67 0.78 0.83 0.66 0.73 0.83 20:25 0.74 0.89 0.90 0.78 0.89 0.93 0.91 14:30 0.63 0.65 0.76 0.82 0.66 0.73 0.82 20:30 0.73 0.89 0.90 0.79 0.90 0.93 0.91 14:35 0.64 0.66 0.77 0.82 0.66 0.74 0.83 20:35 0.74 0.90 0.90 0.79 0.90 0.94 0.91 14:40 0.64 0.66 0.77 0.82 0.66 0.74 0.83 20:40 0.75 0.90 0.90 0.80 0.90 0.94 0.92 14:45 0.64 0.67 0.77 0.81 0.67 0.74 0.84 20:45 0.76 0.91 0.91 0.81 0.91 0.95 0.93 14:50 0.64 0.67 0.77 0.82 0.67 0.75 0.84 20:50 0.76 0.91 0.91 0.81 0.91 0.95 0.93 14:55 0.64 0.68 0.77 0.82 0.67 0.75 0.84 20:55 0.77 0.92 0.91 0.81 0.91 0.95 0.93 15:00 0.62 0.66 0.76 0.78 0.66 0.74 0.83 21:00 0.77 0.92 0.92 0.82 0.91 0.95 0.93 15:05 0.63 0.67 0.76 0.78 0.67 0.74 0.84 21:05 0.78 0.93 0.92 0.82 0.92 0.95 0.93 15:10 0.64 0.68 0.77 0.79 0.67 0.75 0.84 21:10 0.79 0.93 0.92 0.83 0.92 0.95 0.94 15:15 0.64 0.69 0.77 0.79 0.68 0.76 0.85 21:15 0.80 0.94 0.92 0.83 0.93 0.96 0.94 15:20 0.65 0.70 0.78 0.79 0.68 0.77 0.85 21:20 0.81 0.94 0.93 0.84 0.93 0.96 0.95 15:25 0.65 0.70 0.79 0.79 0.69 0.77 0.85 21:25 0.82 0.95 0.93 0.85 0.93 0.96 0.95 15:30 0.64 0.70 0.78 0.78 0.69 0.77 0.85 21:30 0.82 0.95 0.93 0.85 0.93 0.96 0.95 15:35 0.65 0.71 0.79 0.78 0.69 0.78 0.86 21:35 0.83 0.95 0.94 0.86 0.94 0.97 0.96 15:40 0.65 0.71 0.80 0.79 0.70 0.79 0.86 21:40 0.84 0.96 0.94 0.86 0.94 0.97 0.96 15:45 0.65 0.72 0.80 0.78 0.70 0.79 0.86 21:45 0.85 0.96 0.95 0.87 0.95 0.97 0.96 15:50 0.65 0.73 0.81 0.79 0.71 0.80 0.87 21:50 0.85 0.96 0.95 0.87 0.95 0.97 0.96 15:55 0.66 0.73 0.81 0.79 0.71 0.80 0.87 21:55 0.86 0.97 0.95 0.87 0.95 0.97 0.96 16:00 0.63 0.72 0.80 0.76 0.70 0.79 0.85 22:00 0.86 0.97 0.95 0.88 0.95 0.97 0.97 16:05 0.64 0.73 0.81 0.76 0.71 0.79 0.86 22:05 0.86 0.97 0.95 0.88 0.95 0.98 0.97 16:10 0.64 0.73 0.82 0.77 0.71 0.80 0.86 22:10 0.87 0.97 0.96 0.89 0.96 0.98 0.97 16:15 0.64 0.75 0.83 0.77 0.72 0.81 0.87 22:15 0.89 0.98 0.96 0.90 0.96 0.98 0.97 16:20 0.65 0.75 0.83 0.77 0.73 0.81 0.87 22:20 0.90 0.98 0.96 0.90 0.96 0.98 0.98 16:25 0.65 0.76 0.84 0.77 0.73 0.81 0.87 22:25 0.90 0.98 0.97 0.90 0.96 0.98 0.98 16:30 0.64 0.75 0.84 0.75 0.73 0.81 0.87 22:30 0.90 0.98 0.97 0.91 0.97 0.98 0.98 16:35 0.64 0.76 0.84 0.75 0.74 0.82 0.87 22:35 0.91 0.98 0.97 0.92 0.97 0.98 0.98 16:40 0.64 0.76 0.85 0.75 0.75 0.83 0.88 22:40 0.92 0.98 0.97 0.92 0.97 0.98 0.98 16:45 0.64 0.77 0.85 0.74 0.76 0.83 0.88 22:45 0.93 0.98 0.97 0.93 0.97 0.99 0.99 16:50 0.65 0.77 0.86 0.75 0.76 0.83 0.88 22:50 0.93 0.98 0.98 0.93 0.97 0.99 0.99 16:55 0.65 0.78 0.86 0.75 0.77 0.84 0.88 22:55 0.94 0.99 0.98 0.93 0.97 0.99 0.99 17:00 0.64 0.76 0.85 0.72 0.77 0.83 0.87 23:00 0.94 0.98 0.98 0.94 0.97 0.99 0.99 17:05 0.64 0.76 0.85 0.72 0.78 0.83 0.87 23:05 0.94 0.99 0.98 0.95 0.98 0.99 0.99 17:10 0.65 0.77 0.86 0.73 0.79 0.83 0.87 23:10 0.95 0.99 0.99 0.95 0.98 0.99 0.99 17:15 0.65 0.77 0.86 0.73 0.80 0.84 0.88 23:15 0.95 0.99 0.99 0.96 0.98 0.99 0.99 17:20 0.66 0.77 0.86 0.73 0.82 0.84 0.88 23:20 0.96 0.99 0.99 0.96 0.98 0.99 0.99 17:25 0.66 0.77 0.87 0.73 0.82 0.84 0.88 23:25 0.96 0.99 0.99 0.97 0.98 0.99 0.99 17:30 0.65 0.77 0.85 0.72 0.82 0.84 0.87 23:30 0.97 0.99 0.99 0.97 0.99 0.99 0.99 17:35 0.66 0.77 0.86 0.72 0.83 0.84 0.87 23:35 0.97 0.99 0.99 0.97 0.99 0.99 0.99 17:40 0.66 0.78 0.86 0.73 0.83 0.84 0.87 23:40 0.97 0.99 0.99 0.98 0.99 0.99 0.99 17:45 0.66 0.78 0.86 0.73 0.84 0.84 0.87 23:45 0.98 0.99 0.99 0.98 0.99 0.99 0.99 17:50 0.66 0.78 0.87 0.73 0.84 0.84 0.87 23:50 0.98 1.00 0.99 0.98 0.99 0.99 0.99 17:55 0.66 0.79 0.87 0.73 0.84 0.85 0.88 23:55 0.99 1.00 0.99 0.99 0.99 0.99 0.99 89 Table A3. Weekday profiles for Day-work Time Under 25 25-34 35-44 45-54 55-64 65-74 Over 75 Time Under 25 25-34 35-44 45-54 55-64 65-74 Over 75 0:00 0.99 0.96 0.98 0.99 0.99 0.99 1.00 6:00 0.88 0.87 0.81 0.80 0.67 0.75 0.87 0:05 0.99 0.96 0.98 0.99 0.99 0.99 1.00 6:05 0.88 0.87 0.81 0.80 0.66 0.75 0.87 0:10 0.99 0.97 0.98 0.99 0.99 0.99 1.00 6:10 0.87 0.86 0.80 0.79 0.65 0.74 0.86 0:15 0.99 0.97 0.98 0.99 0.99 0.99 1.00 6:15 0.85 0.84 0.79 0.77 0.62 0.72 0.85 0:20 0.99 0.97 0.98 0.99 0.99 0.99 1.00 6:20 0.84 0.83 0.77 0.76 0.61 0.70 0.84 0:25 0.99 0.97 0.98 0.99 0.99 0.99 1.00 6:25 0.83 0.82 0.77 0.75 0.60 0.71 0.84 0:30 0.99 0.97 0.98 0.99 1.00 1.00 1.00 6:30 0.74 0.75 0.71 0.70 0.53 0.64 0.81 0:35 0.99 0.97 0.98 0.99 1.00 1.00 1.00 6:35 0.74 0.75 0.70 0.69 0.53 0.64 0.81 0:40 0.99 0.97 0.98 0.99 1.00 1.00 1.00 6:40 0.72 0.74 0.69 0.68 0.52 0.63 0.80 0:45 0.99 0.98 0.98 0.99 1.00 1.00 1.00 6:45 0.68 0.70 0.66 0.66 0.49 0.60 0.78 0:50 0.99 0.98 0.98 0.99 1.00 1.00 1.00 6:50 0.65 0.68 0.64 0.65 0.48 0.59 0.77 0:55 0.99 0.98 0.98 0.99 1.00 1.00 1.00 6:55 0.63 0.67 0.63 0.64 0.47 0.58 0.77 1:00 0.99 0.98 0.98 0.99 0.99 1.00 1.00 7:00 0.52 0.59 0.55 0.57 0.41 0.49 0.71 1:05 0.99 0.98 0.98 0.99 1.00 1.00 1.00 7:05 0.50 0.57 0.54 0.56 0.40 0.48 0.71 1:10 0.99 0.98 0.98 0.99 1.00 1.00 1.00 7:10 0.47 0.56 0.53 0.55 0.39 0.46 0.70 1:15 1.00 0.98 0.99 0.99 1.00 1.00 1.00 7:15 0.41 0.50 0.49 0.51 0.34 0.42 0.67 1:20 1.00 0.98 0.99 0.99 1.00 1.00 1.00 7:20 0.40 0.50 0.48 0.50 0.34 0.42 0.67 1:25 1.00 0.98 0.99 0.99 1.00 1.00 1.00 7:25 0.39 0.49 0.47 0.49 0.34 0.40 0.66 1:30 1.00 0.98 0.99 0.99 1.00 1.00 1.00 7:30 0.31 0.39 0.40 0.43 0.27 0.32 0.61 1:35 1.00 0.98 0.99 0.99 1.00 1.00 1.00 7:35 0.29 0.38 0.39 0.42 0.26 0.31 0.61 1:40 1.00 0.98 0.99 0.99 1.00 1.00 1.00 7:40 0.26 0.36 0.37 0.41 0.25 0.29 0.60 1:45 1.00 0.98 0.99 0.99 1.00 1.00 1.00 7:45 0.23 0.32 0.34 0.39 0.23 0.27 0.58 1:50 1.00 0.98 0.99 0.99 1.00 1.00 1.00 7:50 0.21 0.30 0.33 0.38 0.22 0.26 0.56 1:55 1.00 0.98 0.99 0.99 1.00 1.00 1.00 7:55 0.20 0.29 0.32 0.37 0.22 0.25 0.56 2:00 1.00 0.98 0.99 0.99 1.00 1.00 1.00 8:00 0.14 0.22 0.26 0.32 0.17 0.20 0.51 2:05 1.00 0.98 0.99 0.99 1.00 1.00 1.00 8:05 0.12 0.20 0.25 0.31 0.16 0.20 0.50 2:10 1.00 0.98 0.99 0.99 1.00 1.00 1.00 8:10 0.12 0.20 0.25 0.31 0.16 0.19 0.50 2:15 1.00 0.98 0.99 0.99 1.00 1.00 1.00 8:15 0.11 0.18 0.23 0.30 0.15 0.18 0.49 2:20 1.00 0.98 0.99 0.99 1.00 1.00 1.00 8:20 0.10 0.17 0.22 0.29 0.14 0.18 0.48 2:25 1.00 0.98 0.99 0.99 1.00 1.00 1.00 8:25 0.10 0.17 0.22 0.29 0.14 0.17 0.48 2:30 1.00 0.98 0.99 0.99 1.00 1.00 1.00 8:30 0.07 0.13 0.19 0.26 0.11 0.14 0.45 2:35 1.00 0.98 0.99 0.99 1.00 1.00 1.00 8:35 0.07 0.12 0.19 0.25 0.11 0.13 0.44 2:40 1.00 0.98 0.99 0.99 1.00 1.00 1.00 8:40 0.06 0.12 0.18 0.25 0.11 0.13 0.44 2:45 1.00 0.98 0.99 0.99 1.00 1.00 1.00 8:45 0.05 0.11 0.17 0.24 0.10 0.11 0.43 2:50 1.00 0.98 0.99 0.99 1.00 1.00 1.00 8:50 0.05 0.10 0.16 0.23 0.09 0.10 0.42 2:55 1.00 0.98 0.99 0.99 1.00 1.00 1.00 8:55 0.05 0.10 0.16 0.23 0.09 0.10 0.42 3:00 1.00 0.98 0.99 0.99 0.99 0.99 1.00 9:00 0.03 0.07 0.14 0.20 0.08 0.08 0.38 3:05 1.00 0.98 0.99 0.99 0.99 0.99 1.00 9:05 0.03 0.07 0.14 0.20 0.07 0.08 0.38 3:10 1.00 0.98 0.99 0.99 0.99 0.99 1.00 9:10 0.03 0.07 0.14 0.20 0.07 0.08 0.37 3:15 1.00 0.98 0.99 0.99 0.99 0.99 1.00 9:15 0.02 0.06 0.13 0.19 0.07 0.07 0.36 3:20 1.00 0.98 0.99 0.99 0.99 0.99 1.00 9:20 0.02 0.06 0.13 0.19 0.07 0.07 0.36 3:25 1.00 0.98 0.99 0.99 0.99 0.99 0.99 9:25 0.02 0.06 0.13 0.19 0.06 0.07 0.36 3:30 1.00 0.98 0.99 0.99 0.99 0.99 0.99 9:30 0.01 0.05 0.12 0.18 0.05 0.06 0.32 3:35 1.00 0.98 0.99 0.99 0.99 0.99 0.99 9:35 0.01 0.04 0.12 0.17 0.05 0.06 0.32 3:40 1.00 0.98 0.98 0.99 0.99 0.99 0.99 9:40 0.01 0.04 0.12 0.17 0.05 0.06 0.31 3:45 1.00 0.98 0.98 0.99 0.99 0.99 0.99 9:45 0.01 0.04 0.11 0.17 0.05 0.06 0.30 3:50 1.00 0.98 0.98 0.99 0.99 0.99 0.99 9:50 0.01 0.04 0.11 0.16 0.04 0.05 0.29 3:55 1.00 0.98 0.98 0.99 0.98 0.99 0.99 9:55 0.01 0.04 0.11 0.16 0.04 0.05 0.29 4:00 0.98 0.99 0.98 0.98 0.97 0.96 0.97 10:00 0.01 0.03 0.10 0.14 0.03 0.05 0.24 4:05 0.98 0.99 0.98 0.98 0.97 0.96 0.97 10:05 0.01 0.03 0.10 0.14 0.03 0.05 0.24 4:10 0.98 0.99 0.98 0.98 0.96 0.96 0.97 10:10 0.01 0.02 0.10 0.14 0.03 0.05 0.24 4:15 0.98 0.99 0.97 0.98 0.96 0.95 0.97 10:15 0.01 0.02 0.10 0.13 0.03 0.05 0.23 4:20 0.98 0.99 0.97 0.98 0.96 0.95 0.97 10:20 0.01 0.02 0.10 0.13 0.03 0.05 0.22 4:25 0.98 0.99 0.97 0.97 0.95 0.95 0.97 10:25 0.01 0.02 0.10 0.13 0.03 0.05 0.22 4:30 0.98 0.99 0.96 0.96 0.94 0.93 0.96 10:30 0.01 0.02 0.09 0.12 0.02 0.05 0.20 4:35 0.98 0.99 0.96 0.96 0.93 0.93 0.96 10:35 0.01 0.02 0.09 0.12 0.02 0.05 0.19 4:40 0.97 0.99 0.96 0.96 0.93 0.93 0.96 10:40 0.01 0.02 0.09 0.12 0.02 0.05 0.19 4:45 0.97 0.98 0.95 0.95 0.91 0.92 0.96 10:45 0.01 0.01 0.09 0.12 0.02 0.05 0.18 4:50 0.97 0.98 0.95 0.95 0.91 0.92 0.96 10:50 0.01 0.01 0.09 0.12 0.02 0.05 0.18 4:55 0.97 0.98 0.95 0.95 0.91 0.91 0.96 10:55 0.01 0.01 0.09 0.12 0.02 0.05 0.18 5:00 0.96 0.98 0.93 0.92 0.88 0.89 0.95 11:00 0.01 0.01 0.08 0.11 0.01 0.05 0.14 5:05 0.96 0.98 0.93 0.92 0.87 0.88 0.94 11:05 0.01 0.01 0.09 0.11 0.01 0.05 0.14 5:10 0.96 0.98 0.92 0.92 0.87 0.88 0.94 11:10 0.01 0.01 0.09 0.11 0.01 0.04 0.13 5:15 0.96 0.97 0.91 0.91 0.85 0.87 0.94 11:15 0.01 0.01 0.08 0.11 0.02 0.05 0.13 5:20 0.95 0.97 0.91 0.90 0.84 0.86 0.93 11:20 0.01 0.01 0.09 0.11 0.01 0.05 0.12 5:25 0.95 0.97 0.91 0.90 0.83 0.86 0.93 11:25 0.01 0.01 0.08 0.11 0.01 0.04 0.12 5:30 0.94 0.95 0.88 0.88 0.78 0.83 0.91 11:30 0.01 0.01 0.08 0.11 0.02 0.05 0.11 5:35 0.93 0.94 0.88 0.87 0.78 0.83 0.91 11:35 0.01 0.01 0.08 0.11 0.02 0.04 0.11 5:40 0.93 0.94 0.88 0.87 0.77 0.82 0.91 11:40 0.02 0.01 0.09 0.11 0.02 0.05 0.11 5:45 0.93 0.93 0.86 0.85 0.75 0.81 0.90 11:45 0.02 0.01 0.08 0.11 0.02 0.05 0.10 5:50 0.92 0.92 0.86 0.85 0.73 0.80 0.90 11:50 0.02 0.01 0.09 0.11 0.02 0.06 0.11 5:55 0.92 0.92 0.85 0.84 0.72 0.80 0.90 11:55 0.02 0.01 0.09 0.11 0.02 0.06 0.12 90 Table A3(cont’d) Time Under 25 25-34 35-44 45-54 55-64 65-74 Over 75 Time Under 25 25-34 35-44 45-54 55-64 65-74 Over 75 12:00 0.02 0.02 0.09 0.11 0.03 0.07 0.11 18:00 0.57 0.44 0.46 0.59 0.43 0.59 0.72 12:05 0.02 0.02 0.09 0.12 0.03 0.08 0.12 18:05 0.59 0.45 0.48 0.60 0.44 0.60 0.73 12:10 0.02 0.03 0.10 0.12 0.03 0.08 0.12 18:10 0.59 0.47 0.49 0.61 0.45 0.61 0.75 12:15 0.03 0.03 0.10 0.12 0.04 0.08 0.12 18:15 0.61 0.49 0.50 0.63 0.48 0.62 0.76 12:20 0.03 0.03 0.10 0.12 0.03 0.08 0.12 18:20 0.62 0.50 0.51 0.64 0.49 0.62 0.77 12:25 0.03 0.03 0.10 0.12 0.03 0.08 0.12 18:25 0.62 0.51 0.52 0.64 0.50 0.63 0.78 12:30 0.02 0.03 0.09 0.12 0.03 0.08 0.12 18:30 0.63 0.54 0.54 0.66 0.53 0.63 0.79 12:35 0.02 0.03 0.10 0.13 0.03 0.08 0.13 18:35 0.64 0.55 0.54 0.66 0.54 0.64 0.80 12:40 0.03 0.03 0.10 0.13 0.03 0.08 0.13 18:40 0.64 0.55 0.55 0.67 0.55 0.64 0.81 12:45 0.02 0.03 0.10 0.13 0.03 0.08 0.12 18:45 0.65 0.56 0.56 0.68 0.56 0.64 0.82 12:50 0.02 0.02 0.09 0.13 0.03 0.09 0.13 18:50 0.65 0.57 0.56 0.69 0.57 0.64 0.83 12:55 0.02 0.02 0.09 0.13 0.02 0.09 0.13 18:55 0.66 0.58 0.57 0.69 0.57 0.64 0.83 13:00 0.02 0.02 0.08 0.12 0.02 0.09 0.13 19:00 0.66 0.58 0.58 0.70 0.59 0.65 0.85 13:05 0.03 0.02 0.08 0.13 0.02 0.09 0.14 19:05 0.67 0.59 0.60 0.71 0.60 0.65 0.86 13:10 0.02 0.02 0.08 0.13 0.02 0.09 0.14 19:10 0.68 0.61 0.61 0.72 0.61 0.64 0.87 13:15 0.03 0.02 0.08 0.13 0.02 0.10 0.14 19:15 0.69 0.62 0.63 0.73 0.64 0.65 0.89 13:20 0.03 0.02 0.08 0.13 0.02 0.10 0.15 19:20 0.69 0.62 0.63 0.74 0.64 0.65 0.89 13:25 0.02 0.01 0.08 0.14 0.02 0.11 0.15 19:25 0.70 0.63 0.64 0.74 0.64 0.65 0.89 13:30 0.03 0.01 0.08 0.14 0.02 0.11 0.15 19:30 0.71 0.64 0.65 0.75 0.66 0.65 0.90 13:35 0.03 0.01 0.08 0.14 0.02 0.13 0.15 19:35 0.71 0.65 0.66 0.76 0.67 0.65 0.91 13:40 0.03 0.01 0.08 0.14 0.02 0.13 0.15 19:40 0.72 0.66 0.67 0.77 0.68 0.65 0.91 13:45 0.03 0.01 0.08 0.14 0.02 0.13 0.15 19:45 0.72 0.67 0.68 0.78 0.69 0.67 0.92 13:50 0.03 0.01 0.08 0.15 0.02 0.13 0.16 19:50 0.73 0.68 0.69 0.78 0.70 0.67 0.93 13:55 0.03 0.01 0.08 0.15 0.01 0.13 0.16 19:55 0.73 0.69 0.70 0.79 0.70 0.68 0.93 14:00 0.04 0.02 0.08 0.15 0.02 0.14 0.15 20:00 0.75 0.71 0.72 0.80 0.72 0.69 0.94 14:05 0.05 0.02 0.08 0.15 0.02 0.14 0.15 20:05 0.75 0.71 0.72 0.80 0.73 0.69 0.94 14:10 0.05 0.02 0.08 0.15 0.02 0.14 0.16 20:10 0.76 0.72 0.73 0.81 0.74 0.69 0.95 14:15 0.06 0.02 0.08 0.15 0.03 0.15 0.17 20:15 0.77 0.73 0.74 0.82 0.75 0.71 0.95 14:20 0.07 0.02 0.08 0.15 0.03 0.15 0.17 20:20 0.77 0.74 0.75 0.83 0.75 0.71 0.95 14:25 0.08 0.02 0.09 0.16 0.03 0.16 0.17 20:25 0.78 0.74 0.76 0.83 0.76 0.72 0.96 14:30 0.09 0.02 0.08 0.16 0.03 0.15 0.17 20:30 0.78 0.75 0.78 0.84 0.78 0.73 0.96 14:35 0.12 0.02 0.09 0.16 0.04 0.16 0.18 20:35 0.79 0.76 0.78 0.85 0.79 0.74 0.97 14:40 0.12 0.02 0.09 0.16 0.04 0.16 0.18 20:40 0.79 0.77 0.79 0.85 0.79 0.75 0.97 14:45 0.13 0.02 0.09 0.17 0.05 0.16 0.18 20:45 0.81 0.78 0.81 0.87 0.81 0.77 0.98 14:50 0.15 0.02 0.10 0.17 0.05 0.16 0.19 20:50 0.81 0.78 0.81 0.87 0.81 0.77 0.98 14:55 0.16 0.02 0.10 0.17 0.05 0.17 0.20 20:55 0.81 0.79 0.82 0.87 0.82 0.78 0.98 15:00 0.18 0.02 0.10 0.18 0.06 0.17 0.20 21:00 0.82 0.80 0.83 0.88 0.83 0.79 0.98 15:05 0.20 0.03 0.10 0.18 0.07 0.17 0.20 21:05 0.83 0.80 0.84 0.88 0.84 0.80 0.98 15:10 0.22 0.03 0.11 0.19 0.08 0.17 0.21 21:10 0.84 0.81 0.84 0.89 0.85 0.81 0.98 15:15 0.24 0.04 0.12 0.20 0.08 0.17 0.22 21:15 0.86 0.82 0.85 0.90 0.86 0.83 0.98 15:20 0.27 0.04 0.13 0.22 0.09 0.17 0.24 21:20 0.86 0.83 0.86 0.91 0.87 0.84 0.98 15:25 0.27 0.04 0.13 0.22 0.09 0.17 0.24 21:25 0.87 0.83 0.86 0.91 0.87 0.84 0.98 15:30 0.30 0.05 0.13 0.23 0.10 0.18 0.26 21:30 0.87 0.84 0.87 0.92 0.88 0.85 0.98 15:35 0.32 0.06 0.14 0.24 0.10 0.19 0.27 21:35 0.88 0.85 0.88 0.92 0.89 0.86 0.98 15:40 0.34 0.06 0.15 0.25 0.11 0.19 0.29 21:40 0.88 0.85 0.88 0.93 0.89 0.87 0.99 15:45 0.35 0.07 0.15 0.26 0.12 0.20 0.30 21:45 0.88 0.86 0.89 0.93 0.90 0.88 0.99 15:50 0.36 0.08 0.16 0.27 0.13 0.20 0.32 21:50 0.89 0.87 0.89 0.94 0.91 0.89 0.99 15:55 0.37 0.08 0.17 0.27 0.14 0.21 0.33 21:55 0.90 0.87 0.90 0.94 0.91 0.89 0.99 16:00 0.38 0.09 0.18 0.29 0.15 0.21 0.35 22:00 0.90 0.88 0.90 0.94 0.92 0.91 0.99 16:05 0.39 0.10 0.18 0.30 0.16 0.23 0.37 22:05 0.91 0.89 0.91 0.95 0.93 0.92 0.99 16:10 0.39 0.11 0.19 0.31 0.17 0.24 0.38 22:10 0.91 0.90 0.91 0.95 0.93 0.92 0.99 16:15 0.42 0.12 0.21 0.34 0.19 0.26 0.42 22:15 0.92 0.90 0.92 0.96 0.94 0.93 0.99 16:20 0.42 0.12 0.21 0.34 0.19 0.27 0.42 22:20 0.93 0.91 0.92 0.96 0.94 0.94 1.00 16:25 0.43 0.13 0.22 0.35 0.20 0.28 0.44 22:25 0.93 0.91 0.93 0.96 0.95 0.94 0.99 16:30 0.44 0.14 0.23 0.36 0.22 0.29 0.46 22:30 0.94 0.91 0.93 0.96 0.96 0.95 1.00 16:35 0.45 0.15 0.24 0.37 0.22 0.30 0.47 22:35 0.94 0.92 0.93 0.97 0.96 0.96 1.00 16:40 0.46 0.16 0.25 0.38 0.24 0.31 0.48 22:40 0.94 0.92 0.93 0.97 0.96 0.96 1.00 16:45 0.46 0.18 0.27 0.40 0.24 0.32 0.50 22:45 0.95 0.92 0.94 0.97 0.97 0.96 1.00 16:50 0.47 0.19 0.27 0.41 0.25 0.34 0.51 22:50 0.95 0.92 0.94 0.97 0.97 0.97 1.00 16:55 0.48 0.19 0.28 0.41 0.25 0.35 0.52 22:55 0.95 0.93 0.94 0.97 0.97 0.97 1.00 17:00 0.49 0.23 0.30 0.45 0.27 0.41 0.56 23:00 0.96 0.94 0.95 0.97 0.98 0.97 1.00 17:05 0.49 0.24 0.30 0.45 0.27 0.41 0.56 23:05 0.96 0.94 0.95 0.98 0.98 0.98 1.00 17:10 0.50 0.26 0.32 0.46 0.29 0.42 0.57 23:10 0.97 0.94 0.95 0.98 0.98 0.98 1.00 17:15 0.51 0.29 0.33 0.48 0.29 0.45 0.59 23:15 0.97 0.95 0.96 0.98 0.98 0.98 1.00 17:20 0.53 0.30 0.35 0.50 0.30 0.46 0.60 23:20 0.97 0.95 0.96 0.98 0.98 0.99 1.00 17:25 0.53 0.31 0.36 0.50 0.31 0.48 0.61 23:25 0.97 0.95 0.96 0.98 0.98 0.99 1.00 17:30 0.54 0.35 0.39 0.52 0.34 0.51 0.64 23:30 0.98 0.95 0.97 0.98 0.99 0.99 1.00 17:35 0.55 0.36 0.40 0.53 0.35 0.52 0.65 23:35 0.98 0.96 0.97 0.99 0.99 0.99 1.00 17:40 0.56 0.38 0.41 0.55 0.35 0.54 0.66 23:40 0.98 0.96 0.97 0.99 0.99 0.99 1.00 17:45 0.56 0.41 0.43 0.57 0.38 0.56 0.69 23:45 0.99 0.96 0.97 0.99 0.99 0.99 1.00 17:50 0.56 0.41 0.43 0.57 0.38 0.56 0.69 23:50 0.99 0.96 0.97 0.99 0.99 0.99 1.00 17:55 0.57 0.43 0.44 0.58 0.39 0.57 0.70 23:55 0.99 0.97 0.98 0.99 0.99 0.99 1.00 91 Table A4. Weekday profiles for Night-work Time Under 25 25-34 35-44 45-54 55-64 65-74 Over 75 Time Under 25 25-34 35-44 45-54 55-64 65-74 Over 75 0:00 0.24 0.48 0.12 0.13 0.09 0.02 0.01 6:00 0.73 0.77 0.53 0.54 0.62 0.83 0.87 0:05 0.24 0.48 0.12 0.13 0.09 0.02 0.01 6:05 0.72 0.78 0.54 0.55 0.63 0.83 0.87 0:10 0.24 0.48 0.11 0.13 0.09 0.02 0.01 6:10 0.73 0.78 0.55 0.56 0.63 0.82 0.87 0:15 0.23 0.49 0.11 0.13 0.09 0.02 0.02 6:15 0.73 0.79 0.56 0.57 0.64 0.82 0.87 0:20 0.23 0.50 0.11 0.13 0.09 0.02 0.02 6:20 0.73 0.80 0.57 0.58 0.64 0.82 0.87 0:25 0.24 0.50 0.11 0.13 0.09 0.02 0.02 6:25 0.73 0.80 0.57 0.57 0.64 0.82 0.87 0:30 0.24 0.50 0.11 0.12 0.09 0.02 0.02 6:30 0.73 0.82 0.57 0.58 0.68 0.80 0.85 0:35 0.24 0.50 0.10 0.12 0.09 0.02 0.02 6:35 0.73 0.82 0.58 0.58 0.68 0.80 0.85 0:40 0.24 0.50 0.10 0.11 0.09 0.02 0.02 6:40 0.73 0.82 0.60 0.59 0.68 0.82 0.86 0:45 0.24 0.51 0.10 0.10 0.09 0.02 0.02 6:45 0.72 0.83 0.60 0.59 0.68 0.81 0.85 0:50 0.24 0.51 0.09 0.11 0.09 0.02 0.02 6:50 0.71 0.83 0.61 0.59 0.68 0.81 0.85 0:55 0.24 0.51 0.09 0.11 0.09 0.02 0.02 6:55 0.72 0.83 0.61 0.59 0.67 0.81 0.85 1:00 0.23 0.52 0.08 0.10 0.08 0.02 0.01 7:00 0.69 0.83 0.63 0.60 0.65 0.79 0.82 1:05 0.24 0.52 0.07 0.10 0.08 0.02 0.01 7:05 0.70 0.83 0.64 0.62 0.66 0.79 0.82 1:10 0.24 0.53 0.07 0.10 0.08 0.02 0.01 7:10 0.73 0.84 0.66 0.63 0.67 0.81 0.84 1:15 0.24 0.53 0.07 0.11 0.08 0.02 0.01 7:15 0.73 0.85 0.68 0.64 0.69 0.80 0.83 1:20 0.23 0.53 0.07 0.11 0.08 0.02 0.01 7:20 0.73 0.85 0.69 0.65 0.69 0.81 0.84 1:25 0.24 0.53 0.06 0.11 0.08 0.02 0.01 7:25 0.73 0.86 0.71 0.65 0.69 0.80 0.83 1:30 0.24 0.54 0.05 0.11 0.09 0.02 0.01 7:30 0.71 0.86 0.70 0.66 0.67 0.78 0.81 1:35 0.24 0.54 0.05 0.11 0.10 0.02 0.01 7:35 0.73 0.87 0.70 0.66 0.67 0.79 0.81 1:40 0.24 0.54 0.05 0.11 0.10 0.02 0.01 7:40 0.75 0.86 0.71 0.68 0.69 0.78 0.81 1:45 0.24 0.54 0.06 0.11 0.10 0.03 0.01 7:45 0.74 0.87 0.72 0.68 0.69 0.77 0.80 1:50 0.24 0.55 0.06 0.11 0.10 0.03 0.01 7:50 0.75 0.87 0.74 0.70 0.68 0.77 0.78 1:55 0.25 0.55 0.06 0.11 0.10 0.03 0.01 7:55 0.75 0.88 0.75 0.70 0.69 0.75 0.77 2:00 0.25 0.54 0.04 0.11 0.10 0.03 0.01 8:00 0.73 0.86 0.76 0.70 0.67 0.73 0.74 2:05 0.27 0.55 0.04 0.11 0.11 0.04 0.02 8:05 0.73 0.87 0.78 0.72 0.68 0.73 0.74 2:10 0.28 0.55 0.04 0.11 0.11 0.04 0.02 8:10 0.73 0.87 0.79 0.73 0.68 0.73 0.73 2:15 0.30 0.56 0.04 0.12 0.11 0.04 0.02 8:15 0.72 0.88 0.79 0.73 0.69 0.73 0.73 2:20 0.32 0.57 0.05 0.12 0.12 0.05 0.02 8:20 0.72 0.88 0.79 0.73 0.71 0.72 0.73 2:25 0.32 0.57 0.06 0.12 0.12 0.05 0.02 8:25 0.72 0.88 0.78 0.73 0.70 0.72 0.73 2:30 0.33 0.57 0.06 0.12 0.12 0.06 0.02 8:30 0.70 0.87 0.80 0.73 0.68 0.70 0.70 2:35 0.34 0.57 0.07 0.13 0.12 0.06 0.02 8:35 0.69 0.87 0.82 0.74 0.67 0.70 0.70 2:40 0.35 0.57 0.07 0.13 0.12 0.06 0.02 8:40 0.69 0.87 0.81 0.75 0.68 0.72 0.72 2:45 0.36 0.58 0.07 0.13 0.12 0.06 0.02 8:45 0.69 0.87 0.83 0.77 0.69 0.70 0.70 2:50 0.37 0.58 0.07 0.13 0.12 0.06 0.02 8:50 0.68 0.87 0.82 0.76 0.69 0.70 0.70 2:55 0.36 0.58 0.07 0.13 0.13 0.06 0.02 8:55 0.68 0.87 0.82 0.76 0.69 0.70 0.70 3:00 0.37 0.58 0.07 0.14 0.14 0.09 0.05 9:00 0.67 0.87 0.81 0.75 0.69 0.68 0.69 3:05 0.37 0.59 0.07 0.14 0.15 0.09 0.05 9:05 0.67 0.88 0.81 0.75 0.69 0.69 0.70 3:10 0.38 0.59 0.07 0.14 0.16 0.09 0.05 9:10 0.67 0.87 0.83 0.76 0.69 0.69 0.70 3:15 0.39 0.59 0.08 0.14 0.16 0.09 0.05 9:15 0.68 0.88 0.83 0.76 0.69 0.69 0.70 3:20 0.39 0.60 0.09 0.15 0.16 0.09 0.05 9:20 0.68 0.87 0.83 0.76 0.69 0.68 0.70 3:25 0.39 0.60 0.10 0.15 0.16 0.09 0.05 9:25 0.68 0.88 0.83 0.77 0.69 0.68 0.70 3:30 0.40 0.61 0.11 0.16 0.17 0.09 0.05 9:30 0.65 0.87 0.83 0.76 0.67 0.68 0.68 3:35 0.41 0.61 0.13 0.16 0.17 0.09 0.05 9:35 0.66 0.86 0.83 0.76 0.67 0.67 0.67 3:40 0.41 0.61 0.13 0.16 0.17 0.09 0.05 9:40 0.66 0.86 0.82 0.76 0.67 0.67 0.67 3:45 0.41 0.61 0.13 0.18 0.19 0.09 0.05 9:45 0.65 0.86 0.82 0.76 0.68 0.66 0.67 3:50 0.41 0.62 0.13 0.18 0.19 0.09 0.05 9:50 0.65 0.86 0.83 0.76 0.68 0.66 0.67 3:55 0.42 0.62 0.13 0.18 0.19 0.09 0.05 9:55 0.65 0.86 0.83 0.77 0.68 0.66 0.67 4:00 0.74 0.76 0.49 0.52 0.62 0.85 0.89 10:00 0.64 0.85 0.83 0.77 0.68 0.59 0.62 4:05 0.74 0.76 0.49 0.53 0.62 0.85 0.89 10:05 0.64 0.86 0.83 0.76 0.67 0.59 0.61 4:10 0.74 0.76 0.49 0.53 0.63 0.85 0.89 10:10 0.63 0.86 0.84 0.77 0.68 0.60 0.62 4:15 0.74 0.77 0.49 0.53 0.64 0.85 0.89 10:15 0.63 0.86 0.84 0.76 0.69 0.61 0.63 4:20 0.74 0.77 0.50 0.53 0.63 0.85 0.89 10:20 0.62 0.86 0.84 0.77 0.69 0.62 0.65 4:25 0.74 0.77 0.50 0.53 0.63 0.86 0.90 10:25 0.62 0.86 0.84 0.77 0.69 0.62 0.66 4:30 0.74 0.77 0.50 0.53 0.65 0.86 0.90 10:30 0.61 0.86 0.84 0.77 0.69 0.62 0.66 4:35 0.74 0.77 0.50 0.53 0.65 0.86 0.90 10:35 0.61 0.86 0.84 0.77 0.68 0.61 0.65 4:40 0.75 0.77 0.50 0.53 0.65 0.86 0.90 10:40 0.61 0.86 0.84 0.77 0.69 0.61 0.66 4:45 0.75 0.78 0.50 0.53 0.65 0.86 0.90 10:45 0.61 0.86 0.84 0.77 0.68 0.60 0.65 4:50 0.75 0.78 0.50 0.53 0.65 0.86 0.90 10:50 0.61 0.87 0.85 0.77 0.69 0.60 0.65 4:55 0.75 0.78 0.50 0.53 0.65 0.86 0.90 10:55 0.60 0.87 0.86 0.77 0.69 0.61 0.65 5:00 0.75 0.77 0.51 0.54 0.64 0.86 0.90 11:00 0.56 0.86 0.83 0.77 0.67 0.59 0.63 5:05 0.75 0.77 0.52 0.54 0.64 0.86 0.90 11:05 0.56 0.87 0.83 0.77 0.66 0.57 0.61 5:10 0.75 0.77 0.52 0.55 0.64 0.86 0.90 11:10 0.56 0.87 0.82 0.77 0.66 0.57 0.61 5:15 0.75 0.77 0.52 0.55 0.64 0.87 0.90 11:15 0.57 0.87 0.83 0.77 0.66 0.57 0.62 5:20 0.74 0.78 0.52 0.55 0.64 0.87 0.90 11:20 0.57 0.88 0.83 0.77 0.66 0.58 0.63 5:25 0.75 0.78 0.52 0.55 0.64 0.87 0.90 11:25 0.57 0.88 0.83 0.77 0.66 0.58 0.63 5:30 0.74 0.78 0.52 0.55 0.64 0.85 0.89 11:30 0.56 0.87 0.83 0.76 0.64 0.60 0.64 5:35 0.73 0.78 0.52 0.55 0.64 0.86 0.90 11:35 0.56 0.88 0.83 0.75 0.65 0.60 0.65 5:40 0.73 0.78 0.52 0.55 0.63 0.86 0.90 11:40 0.55 0.87 0.83 0.76 0.66 0.62 0.67 5:45 0.72 0.78 0.53 0.55 0.63 0.85 0.89 11:45 0.55 0.88 0.83 0.75 0.66 0.62 0.67 5:50 0.72 0.78 0.53 0.55 0.63 0.85 0.88 11:50 0.55 0.88 0.84 0.76 0.65 0.62 0.67 5:55 0.72 0.78 0.53 0.55 0.63 0.85 0.88 11:55 0.55 0.88 0.84 0.76 0.65 0.62 0.67 92 Table A4(cont’d) Time Under 25 25-34 35-44 45-54 55-64 65-74 Over 75 Time Under 25 25-34 35-44 45-54 55-64 65-74 Over 75 12:00 0.53 0.84 0.84 0.74 0.65 0.62 0.67 18:00 0.52 0.29 0.62 0.61 0.59 0.69 0.70 12:05 0.53 0.84 0.84 0.75 0.65 0.63 0.68 18:05 0.53 0.29 0.61 0.61 0.60 0.69 0.69 12:10 0.54 0.84 0.84 0.75 0.65 0.64 0.68 18:10 0.53 0.30 0.60 0.61 0.60 0.68 0.69 12:15 0.55 0.84 0.84 0.76 0.63 0.66 0.70 18:15 0.54 0.30 0.60 0.60 0.60 0.68 0.67 12:20 0.56 0.84 0.84 0.76 0.64 0.64 0.69 18:20 0.54 0.30 0.60 0.60 0.60 0.68 0.67 12:25 0.56 0.84 0.84 0.76 0.64 0.64 0.69 18:25 0.53 0.30 0.60 0.60 0.60 0.67 0.67 12:30 0.56 0.81 0.84 0.77 0.63 0.61 0.65 18:30 0.54 0.29 0.58 0.58 0.59 0.68 0.67 12:35 0.56 0.81 0.83 0.76 0.62 0.62 0.66 18:35 0.54 0.28 0.58 0.58 0.59 0.68 0.67 12:40 0.57 0.82 0.84 0.76 0.62 0.62 0.67 18:40 0.53 0.28 0.57 0.59 0.59 0.67 0.66 12:45 0.55 0.81 0.84 0.74 0.63 0.62 0.67 18:45 0.54 0.28 0.57 0.60 0.58 0.67 0.65 12:50 0.55 0.81 0.84 0.74 0.62 0.63 0.68 18:50 0.53 0.28 0.57 0.59 0.58 0.67 0.66 12:55 0.55 0.81 0.83 0.74 0.63 0.63 0.68 18:55 0.53 0.28 0.57 0.58 0.58 0.66 0.65 13:00 0.54 0.78 0.81 0.73 0.62 0.62 0.66 19:00 0.51 0.27 0.54 0.58 0.56 0.62 0.59 13:05 0.55 0.77 0.82 0.74 0.63 0.62 0.67 19:05 0.51 0.27 0.54 0.59 0.56 0.62 0.59 13:10 0.55 0.77 0.82 0.73 0.63 0.62 0.67 19:10 0.52 0.26 0.55 0.59 0.55 0.60 0.57 13:15 0.55 0.76 0.83 0.73 0.63 0.62 0.66 19:15 0.53 0.27 0.54 0.59 0.54 0.59 0.56 13:20 0.56 0.76 0.83 0.73 0.64 0.62 0.67 19:20 0.53 0.28 0.54 0.59 0.54 0.58 0.56 13:25 0.55 0.76 0.84 0.74 0.64 0.62 0.67 19:25 0.54 0.27 0.53 0.59 0.54 0.58 0.56 13:30 0.55 0.72 0.82 0.74 0.64 0.62 0.66 19:30 0.53 0.26 0.54 0.59 0.52 0.56 0.54 13:35 0.55 0.72 0.83 0.74 0.64 0.63 0.66 19:35 0.54 0.26 0.54 0.59 0.52 0.56 0.53 13:40 0.55 0.71 0.84 0.74 0.64 0.63 0.68 19:40 0.54 0.26 0.55 0.59 0.53 0.56 0.53 13:45 0.53 0.71 0.84 0.74 0.64 0.62 0.67 19:45 0.55 0.26 0.55 0.59 0.52 0.56 0.53 13:50 0.54 0.71 0.82 0.74 0.65 0.62 0.67 19:50 0.55 0.26 0.55 0.59 0.53 0.56 0.53 13:55 0.53 0.71 0.83 0.74 0.65 0.62 0.67 19:55 0.55 0.26 0.54 0.58 0.52 0.56 0.52 14:00 0.53 0.66 0.79 0.73 0.63 0.62 0.67 20:00 0.53 0.26 0.52 0.57 0.48 0.53 0.49 14:05 0.54 0.66 0.80 0.71 0.64 0.62 0.67 20:05 0.53 0.26 0.52 0.56 0.48 0.53 0.49 14:10 0.53 0.66 0.81 0.71 0.63 0.62 0.67 20:10 0.53 0.25 0.51 0.56 0.48 0.53 0.48 14:15 0.53 0.65 0.80 0.72 0.63 0.65 0.69 20:15 0.53 0.25 0.52 0.56 0.49 0.55 0.49 14:20 0.53 0.64 0.81 0.72 0.63 0.65 0.69 20:20 0.54 0.24 0.53 0.54 0.48 0.55 0.49 14:25 0.55 0.64 0.81 0.72 0.63 0.67 0.70 20:25 0.53 0.24 0.51 0.54 0.49 0.54 0.49 14:30 0.54 0.60 0.80 0.69 0.61 0.68 0.72 20:30 0.51 0.22 0.50 0.51 0.47 0.52 0.47 14:35 0.56 0.59 0.80 0.70 0.62 0.68 0.72 20:35 0.50 0.22 0.50 0.50 0.47 0.52 0.47 14:40 0.57 0.59 0.81 0.70 0.63 0.68 0.72 20:40 0.50 0.22 0.50 0.50 0.47 0.52 0.47 14:45 0.57 0.59 0.80 0.71 0.64 0.65 0.69 20:45 0.48 0.21 0.47 0.50 0.46 0.52 0.46 14:50 0.56 0.59 0.80 0.70 0.64 0.66 0.69 20:50 0.48 0.21 0.48 0.50 0.46 0.52 0.46 14:55 0.56 0.60 0.81 0.70 0.64 0.66 0.69 20:55 0.47 0.21 0.48 0.49 0.46 0.53 0.46 15:00 0.53 0.56 0.79 0.69 0.63 0.61 0.65 21:00 0.45 0.19 0.43 0.47 0.43 0.46 0.42 15:05 0.53 0.57 0.79 0.69 0.64 0.62 0.66 21:05 0.44 0.19 0.43 0.47 0.42 0.45 0.41 15:10 0.54 0.57 0.78 0.70 0.64 0.62 0.66 21:10 0.44 0.19 0.43 0.47 0.41 0.46 0.41 15:15 0.53 0.56 0.78 0.71 0.65 0.64 0.67 21:15 0.44 0.19 0.42 0.45 0.39 0.43 0.38 15:20 0.53 0.56 0.80 0.71 0.66 0.63 0.66 21:20 0.45 0.19 0.43 0.45 0.38 0.41 0.37 15:25 0.53 0.57 0.81 0.71 0.66 0.64 0.67 21:25 0.43 0.18 0.42 0.45 0.37 0.41 0.37 15:30 0.53 0.54 0.80 0.69 0.64 0.63 0.66 21:30 0.40 0.18 0.37 0.36 0.33 0.37 0.33 15:35 0.53 0.55 0.80 0.70 0.65 0.62 0.66 21:35 0.40 0.18 0.37 0.35 0.33 0.36 0.32 15:40 0.53 0.56 0.80 0.70 0.66 0.64 0.69 21:40 0.41 0.19 0.37 0.34 0.33 0.36 0.32 15:45 0.55 0.55 0.81 0.70 0.67 0.64 0.69 21:45 0.40 0.19 0.36 0.34 0.31 0.33 0.30 15:50 0.54 0.55 0.79 0.68 0.68 0.67 0.70 21:50 0.38 0.19 0.35 0.32 0.31 0.32 0.28 15:55 0.56 0.55 0.80 0.69 0.69 0.68 0.72 21:55 0.37 0.19 0.35 0.32 0.31 0.32 0.28 16:00 0.52 0.52 0.76 0.66 0.66 0.66 0.69 22:00 0.32 0.20 0.31 0.27 0.24 0.21 0.18 16:05 0.53 0.52 0.76 0.66 0.67 0.66 0.70 22:05 0.31 0.20 0.29 0.27 0.24 0.21 0.18 16:10 0.54 0.52 0.76 0.66 0.67 0.66 0.70 22:10 0.31 0.22 0.29 0.26 0.24 0.21 0.17 16:15 0.53 0.51 0.75 0.65 0.67 0.71 0.74 22:15 0.30 0.22 0.26 0.26 0.22 0.18 0.15 16:20 0.53 0.51 0.75 0.65 0.68 0.71 0.74 22:20 0.30 0.24 0.26 0.25 0.20 0.17 0.15 16:25 0.53 0.51 0.75 0.66 0.69 0.72 0.74 22:25 0.31 0.25 0.26 0.24 0.19 0.17 0.15 16:30 0.53 0.48 0.74 0.65 0.68 0.70 0.73 22:30 0.29 0.25 0.21 0.19 0.16 0.14 0.12 16:35 0.52 0.48 0.74 0.65 0.69 0.72 0.74 22:35 0.30 0.25 0.20 0.17 0.17 0.13 0.11 16:40 0.51 0.47 0.74 0.66 0.69 0.74 0.76 22:40 0.29 0.25 0.19 0.17 0.17 0.13 0.11 16:45 0.52 0.46 0.73 0.66 0.69 0.74 0.76 22:45 0.29 0.26 0.18 0.15 0.16 0.11 0.10 16:50 0.53 0.46 0.73 0.67 0.69 0.75 0.77 22:50 0.28 0.28 0.17 0.15 0.16 0.10 0.09 16:55 0.53 0.46 0.73 0.67 0.69 0.75 0.76 22:55 0.29 0.29 0.18 0.15 0.16 0.09 0.09 17:00 0.53 0.41 0.68 0.63 0.65 0.75 0.76 23:00 0.28 0.31 0.17 0.13 0.12 0.03 0.04 17:05 0.53 0.41 0.68 0.63 0.65 0.75 0.77 23:05 0.28 0.32 0.17 0.13 0.13 0.03 0.04 17:10 0.55 0.40 0.69 0.63 0.64 0.75 0.78 23:10 0.29 0.34 0.16 0.13 0.11 0.03 0.03 17:15 0.55 0.39 0.69 0.65 0.64 0.75 0.78 23:15 0.28 0.36 0.16 0.13 0.11 0.03 0.02 17:20 0.55 0.38 0.69 0.65 0.64 0.74 0.75 23:20 0.28 0.37 0.16 0.13 0.11 0.03 0.02 17:25 0.55 0.37 0.68 0.65 0.65 0.76 0.76 23:25 0.28 0.38 0.16 0.13 0.11 0.03 0.02 17:30 0.53 0.34 0.66 0.65 0.64 0.72 0.73 23:30 0.27 0.42 0.15 0.13 0.10 0.02 0.02 17:35 0.54 0.34 0.66 0.65 0.64 0.72 0.73 23:35 0.27 0.44 0.15 0.14 0.11 0.02 0.02 17:40 0.53 0.33 0.64 0.65 0.63 0.72 0.73 23:40 0.27 0.45 0.15 0.14 0.11 0.02 0.01 17:45 0.53 0.32 0.63 0.64 0.62 0.75 0.75 23:45 0.28 0.48 0.15 0.14 0.12 0.02 0.01 17:50 0.53 0.32 0.63 0.64 0.62 0.75 0.75 23:50 0.27 0.47 0.15 0.14 0.12 0.02 0.01 17:55 0.53 0.32 0.63 0.64 0.61 0.75 0.75 23:55 0.28 0.49 0.15 0.14 0.12 0.02 0.01 93 Table A5. Weekend profiles for Stay-home Time Under 25 25-34 35-44 45-54 55-64 65-74 Over 75 Time Under 25 25-34 35-44 45-54 55-64 65-74 Over 75 0:00 0.97 0.98 0.98 0.99 0.99 0.99 1.00 6:00 0.98 0.95 0.98 0.95 0.96 0.99 0.99 0:05 0.97 0.98 0.99 0.99 0.99 0.99 1.00 6:05 0.98 0.95 0.98 0.95 0.95 0.98 0.99 0:10 0.97 0.98 0.99 0.99 0.99 0.99 1.00 6:10 0.98 0.95 0.98 0.95 0.95 0.98 0.99 0:15 0.98 0.99 0.99 0.99 0.99 0.99 1.00 6:15 0.98 0.95 0.98 0.95 0.95 0.98 0.99 0:20 0.98 0.99 0.99 0.99 1.00 0.99 1.00 6:20 0.98 0.95 0.98 0.95 0.95 0.98 0.99 0:25 0.98 0.99 0.99 0.99 1.00 0.99 1.00 6:25 0.98 0.95 0.98 0.95 0.95 0.98 0.99 0:30 0.98 0.99 0.99 0.99 1.00 1.00 1.00 6:30 0.97 0.94 0.97 0.94 0.94 0.98 0.98 0:35 0.98 0.99 0.99 0.99 1.00 1.00 1.00 6:35 0.97 0.94 0.97 0.94 0.94 0.98 0.98 0:40 0.98 0.99 0.99 0.99 1.00 1.00 1.00 6:40 0.97 0.94 0.97 0.94 0.93 0.98 0.98 0:45 0.98 0.99 0.99 0.99 1.00 1.00 1.00 6:45 0.97 0.93 0.97 0.94 0.93 0.97 0.98 0:50 0.99 0.99 0.99 0.99 1.00 1.00 1.00 6:50 0.97 0.93 0.97 0.94 0.93 0.97 0.98 0:55 0.99 0.99 0.99 0.99 1.00 1.00 1.00 6:55 0.97 0.93 0.97 0.94 0.93 0.97 0.98 1:00 0.98 0.99 0.99 0.99 0.99 1.00 1.00 7:00 0.95 0.92 0.95 0.91 0.91 0.95 0.96 1:05 0.98 0.99 0.99 0.99 0.99 1.00 1.00 7:05 0.95 0.92 0.95 0.91 0.91 0.95 0.96 1:10 0.99 0.99 0.99 1.00 1.00 1.00 1.00 7:10 0.95 0.91 0.95 0.91 0.90 0.95 0.96 1:15 0.99 0.99 0.99 1.00 1.00 1.00 1.00 7:15 0.95 0.91 0.95 0.91 0.90 0.95 0.96 1:20 0.99 0.99 0.99 1.00 1.00 1.00 1.00 7:20 0.95 0.91 0.95 0.91 0.90 0.95 0.96 1:25 0.99 0.99 0.99 1.00 1.00 1.00 1.00 7:25 0.95 0.91 0.95 0.91 0.90 0.95 0.96 1:30 0.99 0.99 0.99 1.00 1.00 1.00 1.00 7:30 0.93 0.90 0.94 0.89 0.88 0.94 0.94 1:35 0.99 0.99 0.99 1.00 1.00 1.00 1.00 7:35 0.93 0.90 0.94 0.89 0.88 0.94 0.94 1:40 0.99 0.99 0.99 1.00 1.00 1.00 1.00 7:40 0.93 0.89 0.94 0.88 0.87 0.93 0.94 1:45 0.99 0.99 0.99 1.00 1.00 1.00 1.00 7:45 0.92 0.89 0.94 0.88 0.87 0.93 0.94 1:50 0.99 0.99 0.99 1.00 1.00 1.00 1.00 7:50 0.92 0.89 0.93 0.88 0.86 0.93 0.93 1:55 1.00 0.99 0.99 1.00 1.00 1.00 1.00 7:55 0.92 0.89 0.93 0.87 0.86 0.92 0.93 2:00 0.99 0.99 0.99 1.00 1.00 1.00 1.00 8:00 0.89 0.86 0.90 0.83 0.83 0.90 0.91 2:05 0.99 0.99 0.99 1.00 1.00 1.00 1.00 8:05 0.89 0.86 0.90 0.83 0.83 0.90 0.91 2:10 0.99 0.99 0.99 1.00 1.00 1.00 1.00 8:10 0.89 0.86 0.90 0.83 0.83 0.90 0.91 2:15 0.99 0.99 0.99 1.00 1.00 1.00 1.00 8:15 0.88 0.85 0.89 0.82 0.82 0.89 0.90 2:20 0.99 1.00 0.99 1.00 1.00 1.00 1.00 8:20 0.88 0.85 0.89 0.82 0.82 0.89 0.90 2:25 0.99 1.00 0.99 1.00 1.00 1.00 1.00 8:25 0.88 0.85 0.89 0.82 0.82 0.89 0.90 2:30 0.99 1.00 1.00 1.00 1.00 1.00 1.00 8:30 0.85 0.82 0.87 0.79 0.79 0.86 0.87 2:35 0.99 1.00 1.00 1.00 1.00 1.00 1.00 8:35 0.85 0.82 0.87 0.78 0.79 0.86 0.87 2:40 1.00 1.00 1.00 1.00 1.00 1.00 1.00 8:40 0.85 0.82 0.86 0.78 0.79 0.86 0.87 2:45 1.00 1.00 1.00 1.00 1.00 1.00 1.00 8:45 0.84 0.81 0.85 0.77 0.78 0.85 0.86 2:50 1.00 1.00 1.00 1.00 1.00 1.00 1.00 8:50 0.83 0.81 0.85 0.77 0.78 0.85 0.85 2:55 1.00 1.00 1.00 1.00 1.00 1.00 1.00 8:55 0.83 0.81 0.85 0.77 0.78 0.85 0.85 3:00 0.99 0.99 0.99 1.00 1.00 1.00 1.00 9:00 0.80 0.76 0.80 0.72 0.74 0.80 0.82 3:05 0.99 0.99 0.99 1.00 1.00 1.00 1.00 9:05 0.80 0.76 0.80 0.72 0.74 0.80 0.82 3:10 1.00 1.00 0.99 1.00 1.00 1.00 1.00 9:10 0.79 0.76 0.80 0.72 0.74 0.80 0.82 3:15 1.00 1.00 0.99 1.00 1.00 1.00 1.00 9:15 0.79 0.75 0.79 0.71 0.73 0.79 0.81 3:20 1.00 0.99 0.99 1.00 1.00 1.00 1.00 9:20 0.78 0.75 0.79 0.71 0.73 0.79 0.81 3:25 1.00 0.99 0.99 1.00 1.00 1.00 1.00 9:25 0.78 0.74 0.79 0.71 0.73 0.79 0.81 3:30 1.00 0.99 0.99 0.99 0.99 1.00 1.00 9:30 0.75 0.72 0.75 0.68 0.70 0.77 0.78 3:35 1.00 0.99 0.99 0.99 0.99 1.00 1.00 9:35 0.75 0.72 0.75 0.68 0.70 0.77 0.79 3:40 1.00 0.99 0.99 0.99 0.99 1.00 1.00 9:40 0.75 0.72 0.75 0.68 0.70 0.77 0.79 3:45 1.00 0.99 0.99 0.99 0.99 1.00 1.00 9:45 0.74 0.71 0.74 0.67 0.70 0.76 0.78 3:50 1.00 0.99 0.99 0.99 0.99 1.00 1.00 9:50 0.73 0.71 0.74 0.67 0.70 0.76 0.77 3:55 1.00 0.99 0.99 0.99 0.99 1.00 1.00 9:55 0.74 0.71 0.74 0.67 0.70 0.76 0.78 4:00 0.99 0.98 0.99 0.98 0.99 1.00 1.00 10:00 0.70 0.66 0.69 0.63 0.67 0.73 0.75 4:05 0.99 0.98 0.99 0.98 0.99 1.00 1.00 10:05 0.70 0.67 0.70 0.63 0.67 0.73 0.75 4:10 0.99 0.98 0.99 0.98 0.99 1.00 1.00 10:10 0.70 0.67 0.70 0.63 0.67 0.73 0.75 4:15 0.99 0.98 0.99 0.98 0.99 1.00 1.00 10:15 0.70 0.66 0.69 0.63 0.67 0.73 0.75 4:20 0.99 0.98 0.99 0.98 0.99 1.00 1.00 10:20 0.69 0.66 0.69 0.63 0.67 0.73 0.75 4:25 0.99 0.98 0.99 0.98 0.99 1.00 1.00 10:25 0.69 0.66 0.69 0.63 0.67 0.73 0.75 4:30 0.99 0.98 0.99 0.98 0.99 1.00 1.00 10:30 0.66 0.63 0.67 0.61 0.65 0.71 0.73 4:35 0.99 0.98 0.99 0.98 0.99 1.00 1.00 10:35 0.66 0.64 0.67 0.61 0.65 0.71 0.73 4:40 0.99 0.98 0.99 0.98 0.99 1.00 1.00 10:40 0.66 0.64 0.67 0.61 0.66 0.71 0.74 4:45 0.99 0.97 0.99 0.98 0.99 1.00 1.00 10:45 0.66 0.63 0.66 0.61 0.65 0.71 0.73 4:50 0.99 0.97 0.99 0.98 0.99 1.00 1.00 10:50 0.66 0.63 0.66 0.61 0.66 0.71 0.74 4:55 0.99 0.97 0.99 0.98 0.99 1.00 1.00 10:55 0.66 0.63 0.66 0.61 0.66 0.71 0.74 5:00 0.99 0.97 0.98 0.98 0.98 1.00 1.00 11:00 0.62 0.59 0.63 0.58 0.64 0.70 0.72 5:05 0.99 0.97 0.98 0.98 0.98 1.00 1.00 11:05 0.62 0.60 0.64 0.59 0.64 0.70 0.72 5:10 0.99 0.97 0.99 0.98 0.98 1.00 1.00 11:10 0.62 0.60 0.64 0.59 0.65 0.70 0.73 5:15 0.99 0.97 0.99 0.97 0.98 1.00 1.00 11:15 0.62 0.60 0.64 0.59 0.65 0.70 0.73 5:20 0.99 0.97 0.98 0.97 0.98 1.00 1.00 11:20 0.63 0.60 0.64 0.59 0.65 0.71 0.73 5:25 0.99 0.97 0.98 0.97 0.98 1.00 1.00 11:25 0.62 0.60 0.64 0.60 0.65 0.71 0.74 5:30 0.99 0.96 0.98 0.97 0.97 0.99 0.99 11:30 0.61 0.59 0.63 0.58 0.65 0.70 0.73 5:35 0.99 0.96 0.98 0.97 0.97 0.99 0.99 11:35 0.61 0.59 0.64 0.59 0.65 0.70 0.73 5:40 0.99 0.96 0.98 0.97 0.97 0.99 0.99 11:40 0.61 0.59 0.64 0.59 0.65 0.71 0.73 5:45 0.99 0.96 0.98 0.97 0.97 0.99 0.99 11:45 0.61 0.59 0.64 0.59 0.65 0.71 0.74 5:50 0.99 0.96 0.98 0.97 0.97 0.99 0.99 11:50 0.61 0.60 0.64 0.59 0.66 0.71 0.74 5:55 0.99 0.96 0.98 0.96 0.97 0.99 0.99 11:55 0.61 0.60 0.65 0.60 0.66 0.71 0.74 94 Table A5(cont’d) Time Under 25 25-34 35-44 45-54 55-64 65-74 Over 75 Time Under 25 25-34 35-44 45-54 55-64 65-74 Over 75 12:00 0.58 0.57 0.62 0.57 0.64 0.70 0.73 18:00 0.66 0.71 0.73 0.73 0.82 0.83 0.89 12:05 0.59 0.58 0.62 0.58 0.65 0.71 0.74 18:05 0.66 0.72 0.73 0.74 0.83 0.84 0.89 12:10 0.59 0.59 0.63 0.59 0.66 0.72 0.75 18:10 0.67 0.73 0.74 0.75 0.83 0.84 0.90 12:15 0.59 0.59 0.63 0.59 0.66 0.72 0.76 18:15 0.68 0.73 0.75 0.75 0.84 0.84 0.90 12:20 0.59 0.60 0.64 0.60 0.66 0.73 0.77 18:20 0.68 0.74 0.76 0.75 0.84 0.85 0.90 12:25 0.60 0.60 0.64 0.60 0.67 0.73 0.77 18:25 0.69 0.74 0.76 0.76 0.84 0.85 0.90 12:30 0.58 0.59 0.63 0.59 0.66 0.73 0.77 18:30 0.69 0.74 0.76 0.76 0.84 0.85 0.91 12:35 0.59 0.60 0.64 0.59 0.67 0.73 0.77 18:35 0.69 0.74 0.76 0.76 0.84 0.85 0.91 12:40 0.59 0.60 0.65 0.60 0.67 0.74 0.78 18:40 0.69 0.75 0.76 0.77 0.85 0.86 0.91 12:45 0.59 0.60 0.65 0.61 0.67 0.74 0.78 18:45 0.70 0.75 0.77 0.77 0.86 0.86 0.91 12:50 0.60 0.61 0.65 0.61 0.68 0.74 0.78 18:50 0.71 0.76 0.77 0.77 0.86 0.86 0.91 12:55 0.60 0.61 0.66 0.61 0.68 0.74 0.79 18:55 0.71 0.76 0.78 0.78 0.86 0.87 0.92 13:00 0.58 0.59 0.64 0.60 0.67 0.73 0.78 19:00 0.70 0.75 0.77 0.77 0.86 0.87 0.92 13:05 0.58 0.60 0.65 0.61 0.68 0.74 0.78 19:05 0.70 0.76 0.78 0.77 0.87 0.87 0.92 13:10 0.58 0.60 0.65 0.61 0.68 0.74 0.78 19:10 0.71 0.76 0.79 0.78 0.87 0.87 0.93 13:15 0.59 0.61 0.65 0.61 0.68 0.75 0.79 19:15 0.72 0.78 0.80 0.79 0.88 0.88 0.93 13:20 0.59 0.61 0.66 0.62 0.69 0.75 0.79 19:20 0.72 0.78 0.80 0.79 0.89 0.88 0.94 13:25 0.60 0.62 0.67 0.62 0.69 0.76 0.80 19:25 0.73 0.78 0.80 0.80 0.89 0.88 0.94 13:30 0.59 0.61 0.66 0.62 0.69 0.75 0.80 19:30 0.73 0.79 0.81 0.80 0.89 0.89 0.94 13:35 0.59 0.61 0.67 0.62 0.69 0.76 0.80 19:35 0.73 0.79 0.81 0.80 0.89 0.89 0.94 13:40 0.60 0.62 0.67 0.62 0.70 0.76 0.81 19:40 0.74 0.80 0.82 0.81 0.90 0.89 0.95 13:45 0.60 0.62 0.67 0.63 0.70 0.76 0.81 19:45 0.75 0.80 0.82 0.81 0.90 0.89 0.95 13:50 0.60 0.62 0.68 0.63 0.71 0.77 0.82 19:50 0.75 0.81 0.83 0.82 0.91 0.90 0.95 13:55 0.60 0.63 0.68 0.63 0.71 0.77 0.82 19:55 0.75 0.81 0.83 0.83 0.91 0.90 0.95 14:00 0.58 0.61 0.66 0.62 0.70 0.76 0.81 20:00 0.75 0.81 0.84 0.83 0.91 0.90 0.95 14:05 0.59 0.62 0.67 0.62 0.71 0.77 0.81 20:05 0.75 0.82 0.84 0.83 0.91 0.90 0.95 14:10 0.59 0.62 0.68 0.63 0.71 0.77 0.82 20:10 0.76 0.82 0.85 0.83 0.91 0.90 0.95 14:15 0.60 0.63 0.68 0.63 0.72 0.78 0.82 20:15 0.77 0.83 0.85 0.84 0.92 0.91 0.96 14:20 0.60 0.63 0.68 0.64 0.72 0.78 0.83 20:20 0.77 0.84 0.86 0.85 0.92 0.91 0.96 14:25 0.61 0.64 0.69 0.64 0.73 0.79 0.83 20:25 0.78 0.84 0.86 0.85 0.92 0.91 0.96 14:30 0.60 0.63 0.68 0.64 0.73 0.79 0.83 20:30 0.78 0.84 0.87 0.86 0.93 0.92 0.96 14:35 0.60 0.64 0.69 0.65 0.74 0.80 0.84 20:35 0.78 0.85 0.87 0.86 0.93 0.92 0.96 14:40 0.60 0.64 0.69 0.65 0.74 0.80 0.84 20:40 0.79 0.86 0.88 0.87 0.93 0.92 0.97 14:45 0.61 0.64 0.69 0.65 0.74 0.80 0.84 20:45 0.80 0.86 0.89 0.88 0.94 0.93 0.97 14:50 0.61 0.65 0.70 0.66 0.74 0.80 0.84 20:50 0.80 0.86 0.89 0.88 0.94 0.93 0.97 14:55 0.62 0.65 0.70 0.66 0.74 0.81 0.85 20:55 0.81 0.87 0.89 0.88 0.94 0.93 0.97 15:00 0.60 0.63 0.68 0.65 0.73 0.79 0.84 21:00 0.80 0.86 0.89 0.88 0.95 0.93 0.97 15:05 0.60 0.64 0.69 0.65 0.74 0.80 0.84 21:05 0.81 0.87 0.89 0.89 0.95 0.93 0.97 15:10 0.61 0.64 0.69 0.66 0.74 0.81 0.84 21:10 0.82 0.88 0.90 0.90 0.95 0.93 0.97 15:15 0.62 0.65 0.69 0.67 0.75 0.81 0.85 21:15 0.82 0.88 0.90 0.90 0.95 0.94 0.98 15:20 0.62 0.65 0.70 0.68 0.75 0.81 0.85 21:20 0.83 0.89 0.91 0.91 0.96 0.94 0.98 15:25 0.63 0.65 0.70 0.68 0.76 0.82 0.85 21:25 0.84 0.90 0.91 0.91 0.96 0.94 0.98 15:30 0.62 0.65 0.69 0.68 0.75 0.81 0.85 21:30 0.84 0.89 0.92 0.91 0.96 0.94 0.98 15:35 0.63 0.65 0.69 0.68 0.75 0.81 0.85 21:35 0.85 0.90 0.92 0.92 0.96 0.95 0.98 15:40 0.63 0.66 0.70 0.69 0.76 0.82 0.86 21:40 0.85 0.90 0.92 0.92 0.96 0.95 0.98 15:45 0.63 0.66 0.70 0.69 0.76 0.82 0.86 21:45 0.86 0.91 0.93 0.93 0.97 0.95 0.98 15:50 0.64 0.66 0.70 0.69 0.76 0.82 0.86 21:50 0.87 0.91 0.93 0.93 0.97 0.96 0.98 15:55 0.64 0.66 0.70 0.69 0.77 0.82 0.87 21:55 0.87 0.92 0.94 0.93 0.97 0.96 0.98 16:00 0.62 0.65 0.68 0.68 0.75 0.80 0.85 22:00 0.87 0.91 0.94 0.93 0.96 0.96 0.98 16:05 0.62 0.65 0.69 0.68 0.76 0.81 0.86 22:05 0.88 0.92 0.94 0.94 0.97 0.96 0.98 16:10 0.63 0.66 0.69 0.69 0.77 0.82 0.87 22:10 0.89 0.92 0.94 0.94 0.97 0.96 0.99 16:15 0.64 0.67 0.70 0.70 0.78 0.82 0.87 22:15 0.90 0.93 0.95 0.94 0.97 0.97 0.99 16:20 0.64 0.67 0.70 0.70 0.78 0.82 0.87 22:20 0.90 0.93 0.95 0.95 0.97 0.97 0.99 16:25 0.65 0.68 0.70 0.71 0.78 0.83 0.87 22:25 0.90 0.93 0.95 0.95 0.98 0.97 0.99 16:30 0.64 0.67 0.69 0.70 0.78 0.82 0.87 22:30 0.91 0.94 0.96 0.95 0.98 0.97 0.99 16:35 0.64 0.68 0.70 0.71 0.78 0.82 0.87 22:35 0.91 0.94 0.96 0.96 0.98 0.97 0.99 16:40 0.64 0.68 0.70 0.71 0.78 0.83 0.87 22:40 0.92 0.95 0.96 0.96 0.98 0.98 0.99 16:45 0.64 0.69 0.70 0.71 0.79 0.82 0.87 22:45 0.92 0.95 0.97 0.96 0.98 0.98 0.99 16:50 0.65 0.69 0.71 0.71 0.79 0.83 0.87 22:50 0.93 0.95 0.97 0.96 0.98 0.98 0.99 16:55 0.65 0.69 0.71 0.72 0.79 0.83 0.87 22:55 0.93 0.95 0.97 0.97 0.98 0.98 0.99 17:00 0.64 0.69 0.70 0.71 0.78 0.82 0.87 23:00 0.93 0.96 0.97 0.97 0.98 0.98 0.99 17:05 0.64 0.69 0.71 0.71 0.78 0.82 0.87 23:05 0.94 0.96 0.97 0.97 0.98 0.98 0.99 17:10 0.65 0.69 0.72 0.72 0.79 0.83 0.87 23:10 0.94 0.96 0.97 0.97 0.98 0.98 1.00 17:15 0.66 0.70 0.72 0.72 0.80 0.83 0.88 23:15 0.95 0.97 0.98 0.98 0.99 0.99 1.00 17:20 0.66 0.70 0.73 0.73 0.80 0.83 0.88 23:20 0.95 0.97 0.98 0.98 0.99 0.99 1.00 17:25 0.67 0.71 0.73 0.73 0.81 0.84 0.88 23:25 0.96 0.97 0.98 0.98 0.99 0.99 1.00 17:30 0.66 0.71 0.72 0.73 0.81 0.83 0.88 23:30 0.96 0.98 0.98 0.98 0.99 0.99 1.00 17:35 0.67 0.71 0.73 0.73 0.82 0.83 0.89 23:35 0.97 0.98 0.98 0.98 0.99 0.99 1.00 17:40 0.67 0.72 0.73 0.74 0.82 0.83 0.89 23:40 0.97 0.98 0.99 0.98 0.99 0.99 1.00 17:45 0.67 0.72 0.74 0.74 0.83 0.84 0.89 23:45 0.98 0.98 0.99 0.99 0.99 0.99 1.00 17:50 0.67 0.72 0.74 0.74 0.83 0.84 0.89 23:50 0.98 0.99 0.99 0.99 0.99 0.99 1.00 17:55 0.67 0.73 0.74 0.74 0.83 0.84 0.89 23:55 0.99 0.99 0.99 0.99 0.99 0.99 1.00 95 Table A6. Weekend profiles for Day-work Time Under 25 25-34 35-44 45-54 55-64 65-74 Over 75 Time Under 25 25-34 35-44 45-54 55-64 65-74 Over 75 0:00 0.89 0.84 0.95 0.90 0.97 0.99 0.99 6:00 0.31 0.82 0.75 0.56 0.84 0.77 0.83 0:05 0.90 0.84 0.96 0.90 0.97 0.99 0.99 6:05 0.30 0.81 0.74 0.55 0.84 0.77 0.83 0:10 0.90 0.84 0.96 0.90 0.97 0.99 0.99 6:10 0.28 0.81 0.73 0.55 0.84 0.76 0.83 0:15 0.90 0.85 0.97 0.91 0.97 0.99 0.99 6:15 0.26 0.79 0.71 0.52 0.83 0.75 0.80 0:20 0.92 0.85 0.97 0.91 0.97 0.99 0.99 6:20 0.24 0.77 0.70 0.50 0.84 0.74 0.79 0:25 0.92 0.85 0.97 0.91 0.97 1.00 0.99 6:25 0.23 0.76 0.69 0.48 0.84 0.74 0.78 0:30 0.92 0.86 0.97 0.93 0.98 1.00 0.99 6:30 0.19 0.71 0.63 0.36 0.82 0.70 0.76 0:35 0.94 0.87 0.97 0.93 0.98 1.00 0.99 6:35 0.19 0.71 0.63 0.36 0.82 0.70 0.76 0:40 0.93 0.87 0.97 0.93 0.98 1.00 1.00 6:40 0.17 0.70 0.62 0.33 0.82 0.69 0.76 0:45 0.93 0.88 0.98 0.93 0.99 1.00 1.00 6:45 0.16 0.67 0.60 0.30 0.82 0.67 0.75 0:50 0.94 0.88 0.98 0.94 0.99 1.00 1.00 6:50 0.15 0.66 0.59 0.28 0.82 0.66 0.74 0:55 0.94 0.88 0.98 0.94 0.99 1.00 1.00 6:55 0.14 0.65 0.59 0.27 0.82 0.66 0.73 1:00 0.92 0.86 0.98 0.95 0.99 0.99 1.00 7:00 0.06 0.55 0.50 0.13 0.77 0.60 0.66 1:05 0.94 0.87 0.98 0.95 0.99 0.99 1.00 7:05 0.07 0.55 0.50 0.13 0.77 0.60 0.67 1:10 0.94 0.87 0.98 0.95 0.99 0.99 1.00 7:10 0.08 0.54 0.49 0.13 0.77 0.59 0.66 1:15 0.94 0.87 0.99 0.95 0.99 0.99 1.00 7:15 0.05 0.52 0.46 0.08 0.76 0.58 0.63 1:20 0.94 0.87 0.99 0.95 0.99 0.99 1.00 7:20 0.05 0.52 0.46 0.08 0.76 0.58 0.63 1:25 0.94 0.87 0.99 0.95 0.99 0.99 1.00 7:25 0.04 0.52 0.46 0.07 0.75 0.58 0.64 1:30 0.94 0.88 0.99 0.95 0.99 1.00 1.00 7:30 0.03 0.46 0.40 0.06 0.74 0.52 0.60 1:35 0.94 0.88 0.99 0.95 1.00 1.00 1.00 7:35 0.02 0.45 0.39 0.06 0.73 0.51 0.60 1:40 0.94 0.88 0.99 0.95 1.00 1.00 1.00 7:40 0.01 0.43 0.38 0.06 0.73 0.49 0.59 1:45 0.94 0.88 0.99 0.95 1.00 1.00 1.00 7:45 0.02 0.41 0.36 0.06 0.73 0.47 0.58 1:50 0.94 0.89 0.99 0.95 1.00 1.00 1.00 7:50 0.02 0.41 0.34 0.05 0.72 0.46 0.58 1:55 0.95 0.89 0.99 0.95 1.00 1.00 1.00 7:55 0.02 0.40 0.34 0.05 0.72 0.46 0.58 2:00 0.96 0.91 0.99 0.95 1.00 1.00 1.00 8:00 0.02 0.31 0.27 0.05 0.67 0.39 0.54 2:05 0.95 0.91 0.99 0.95 1.00 1.00 1.00 8:05 0.02 0.29 0.25 0.04 0.67 0.37 0.53 2:10 0.95 0.91 0.99 0.95 0.99 1.00 1.00 8:10 0.02 0.29 0.25 0.04 0.66 0.37 0.53 2:15 0.95 0.91 0.99 0.95 0.99 1.00 1.00 8:15 0.01 0.27 0.24 0.04 0.66 0.34 0.52 2:20 0.96 0.92 0.99 0.95 1.00 1.00 1.00 8:20 0.01 0.27 0.23 0.04 0.66 0.34 0.50 2:25 0.96 0.92 0.99 0.95 1.00 1.00 1.00 8:25 0.01 0.27 0.23 0.04 0.66 0.33 0.51 2:30 0.97 0.93 0.99 0.95 0.99 0.99 1.00 8:30 0.01 0.21 0.18 0.03 0.65 0.29 0.47 2:35 0.97 0.93 0.99 0.96 1.00 0.99 1.00 8:35 0.01 0.21 0.18 0.04 0.64 0.29 0.47 2:40 0.97 0.93 0.99 0.96 1.00 0.99 1.00 8:40 0.01 0.21 0.17 0.04 0.64 0.28 0.47 2:45 0.97 0.94 0.99 0.96 1.00 0.99 1.00 8:45 0.01 0.20 0.16 0.03 0.63 0.25 0.44 2:50 0.97 0.94 0.99 0.95 1.00 0.99 1.00 8:50 0.03 0.19 0.15 0.02 0.62 0.24 0.44 2:55 0.97 0.94 0.99 0.95 1.00 0.99 1.00 8:55 0.03 0.19 0.15 0.02 0.62 0.24 0.44 3:00 0.94 0.94 0.99 0.96 0.99 0.99 1.00 9:00 0.03 0.12 0.10 0.01 0.55 0.17 0.38 3:05 0.94 0.94 0.99 0.96 0.99 0.99 1.00 9:05 0.03 0.12 0.10 0.01 0.55 0.16 0.38 3:10 0.94 0.95 0.99 0.96 0.99 0.99 1.00 9:10 0.03 0.11 0.10 0.01 0.55 0.16 0.37 3:15 0.94 0.94 0.99 0.96 1.00 0.99 1.00 9:15 0.03 0.11 0.09 0.01 0.54 0.14 0.35 3:20 0.94 0.95 0.99 0.96 1.00 0.99 1.00 9:20 0.03 0.10 0.09 0.01 0.54 0.14 0.36 3:25 0.94 0.95 0.99 0.96 1.00 0.99 1.00 9:25 0.03 0.09 0.09 0.01 0.54 0.13 0.35 3:30 0.96 0.95 0.99 0.96 0.99 0.99 1.00 9:30 0.03 0.08 0.07 0.02 0.50 0.11 0.29 3:35 0.97 0.94 0.99 0.96 0.99 0.99 1.00 9:35 0.03 0.07 0.07 0.02 0.49 0.10 0.30 3:40 0.97 0.94 0.99 0.96 0.99 0.99 1.00 9:40 0.02 0.06 0.07 0.02 0.49 0.10 0.30 3:45 0.96 0.94 0.99 0.95 0.99 0.99 1.00 9:45 0.02 0.06 0.07 0.03 0.47 0.09 0.29 3:50 0.94 0.94 0.99 0.96 0.99 0.99 1.00 9:50 0.02 0.06 0.07 0.02 0.46 0.09 0.28 3:55 0.94 0.94 0.99 0.95 0.99 0.99 1.00 9:55 0.02 0.06 0.07 0.02 0.45 0.09 0.26 4:00 0.72 0.96 0.95 0.96 0.97 0.94 0.96 10:00 0.02 0.03 0.06 0.02 0.36 0.06 0.22 4:05 0.72 0.96 0.95 0.96 0.97 0.94 0.96 10:05 0.02 0.03 0.06 0.02 0.35 0.07 0.22 4:10 0.70 0.96 0.95 0.95 0.97 0.94 0.96 10:10 0.02 0.03 0.06 0.02 0.35 0.07 0.21 4:15 0.70 0.95 0.95 0.95 0.96 0.94 0.96 10:15 0.02 0.03 0.06 0.02 0.33 0.07 0.21 4:20 0.70 0.95 0.95 0.96 0.96 0.94 0.96 10:20 0.01 0.02 0.06 0.02 0.33 0.07 0.22 4:25 0.69 0.95 0.94 0.96 0.96 0.94 0.95 10:25 0.01 0.02 0.06 0.02 0.33 0.08 0.22 4:30 0.63 0.94 0.93 0.94 0.95 0.94 0.95 10:30 0.01 0.03 0.06 0.02 0.28 0.08 0.20 4:35 0.61 0.94 0.93 0.93 0.95 0.94 0.95 10:35 0.02 0.04 0.06 0.02 0.28 0.08 0.21 4:40 0.59 0.94 0.92 0.93 0.95 0.93 0.95 10:40 0.03 0.03 0.06 0.02 0.27 0.08 0.20 4:45 0.57 0.94 0.92 0.91 0.94 0.92 0.95 10:45 0.04 0.02 0.06 0.01 0.26 0.08 0.22 4:50 0.57 0.94 0.92 0.91 0.94 0.92 0.95 10:50 0.05 0.02 0.06 0.02 0.26 0.09 0.22 4:55 0.57 0.94 0.92 0.91 0.93 0.92 0.95 10:55 0.05 0.01 0.07 0.02 0.25 0.09 0.22 5:00 0.52 0.94 0.88 0.87 0.92 0.90 0.92 11:00 0.04 0.01 0.06 0.01 0.20 0.10 0.20 5:05 0.52 0.94 0.88 0.86 0.92 0.90 0.92 11:05 0.06 0.01 0.06 0.01 0.20 0.10 0.20 5:10 0.51 0.94 0.88 0.85 0.92 0.89 0.92 11:10 0.08 0.02 0.07 0.01 0.20 0.10 0.20 5:15 0.50 0.94 0.87 0.83 0.91 0.87 0.91 11:15 0.08 0.02 0.07 0.02 0.20 0.10 0.20 5:20 0.49 0.93 0.86 0.82 0.91 0.86 0.91 11:20 0.08 0.02 0.07 0.01 0.20 0.10 0.20 5:25 0.48 0.93 0.86 0.81 0.90 0.87 0.91 11:25 0.08 0.02 0.07 0.01 0.19 0.10 0.21 5:30 0.43 0.91 0.83 0.75 0.89 0.84 0.90 11:30 0.08 0.01 0.08 0.02 0.18 0.09 0.18 5:35 0.41 0.91 0.83 0.74 0.89 0.84 0.90 11:35 0.08 0.02 0.08 0.01 0.17 0.09 0.18 5:40 0.40 0.91 0.82 0.74 0.89 0.84 0.90 11:40 0.08 0.01 0.08 0.02 0.18 0.10 0.19 5:45 0.39 0.89 0.81 0.71 0.88 0.83 0.88 11:45 0.09 0.01 0.08 0.02 0.17 0.10 0.20 5:50 0.37 0.89 0.80 0.70 0.87 0.83 0.87 11:50 0.10 0.01 0.08 0.02 0.17 0.10 0.20 5:55 0.36 0.88 0.80 0.70 0.87 0.83 0.87 11:55 0.10 0.01 0.08 0.02 0.17 0.10 0.20 96 Table A6(cont’d) Time Under 25 25-34 35-44 45-54 55-64 65-74 Over 75 Time Under 25 25-34 35-44 45-54 55-64 65-74 Over 75 12:00 0.10 0.01 0.08 0.02 0.16 0.09 0.18 18:00 0.32 0.09 0.53 0.26 0.17 0.48 0.30 12:05 0.10 0.01 0.09 0.03 0.17 0.09 0.18 18:05 0.32 0.10 0.53 0.30 0.18 0.49 0.29 12:10 0.10 0.01 0.09 0.02 0.17 0.09 0.18 18:10 0.34 0.10 0.54 0.31 0.19 0.50 0.28 12:15 0.10 0.02 0.10 0.03 0.17 0.09 0.20 18:15 0.37 0.11 0.55 0.32 0.19 0.51 0.28 12:20 0.10 0.02 0.10 0.02 0.17 0.09 0.20 18:20 0.38 0.11 0.55 0.34 0.20 0.53 0.27 12:25 0.10 0.02 0.11 0.02 0.17 0.09 0.20 18:25 0.39 0.11 0.55 0.35 0.20 0.53 0.27 12:30 0.10 0.04 0.11 0.03 0.17 0.08 0.21 18:30 0.41 0.08 0.56 0.36 0.22 0.56 0.29 12:35 0.10 0.04 0.10 0.03 0.17 0.07 0.21 18:35 0.42 0.09 0.56 0.37 0.21 0.56 0.29 12:40 0.10 0.04 0.11 0.03 0.16 0.08 0.21 18:40 0.44 0.08 0.56 0.38 0.21 0.56 0.30 12:45 0.09 0.04 0.10 0.02 0.16 0.08 0.22 18:45 0.46 0.08 0.57 0.38 0.22 0.57 0.31 12:50 0.10 0.04 0.10 0.03 0.17 0.08 0.24 18:50 0.48 0.08 0.57 0.39 0.23 0.59 0.32 12:55 0.10 0.04 0.10 0.03 0.16 0.09 0.24 18:55 0.48 0.08 0.57 0.38 0.23 0.60 0.31 13:00 0.10 0.04 0.10 0.04 0.14 0.08 0.24 19:00 0.47 0.07 0.57 0.43 0.23 0.62 0.31 13:05 0.11 0.04 0.10 0.03 0.15 0.09 0.26 19:05 0.46 0.07 0.57 0.42 0.22 0.63 0.32 13:10 0.12 0.04 0.11 0.03 0.15 0.09 0.26 19:10 0.48 0.06 0.57 0.43 0.22 0.63 0.32 13:15 0.14 0.05 0.11 0.03 0.16 0.08 0.28 19:15 0.52 0.06 0.59 0.45 0.24 0.67 0.32 13:20 0.14 0.06 0.10 0.03 0.16 0.08 0.27 19:20 0.52 0.06 0.59 0.45 0.24 0.67 0.32 13:25 0.14 0.06 0.11 0.03 0.16 0.09 0.26 19:25 0.52 0.06 0.59 0.46 0.25 0.68 0.33 13:30 0.15 0.07 0.10 0.03 0.17 0.10 0.26 19:30 0.52 0.07 0.60 0.49 0.25 0.69 0.32 13:35 0.15 0.08 0.10 0.03 0.17 0.10 0.26 19:35 0.52 0.07 0.60 0.51 0.26 0.69 0.33 13:40 0.16 0.07 0.10 0.03 0.18 0.11 0.27 19:40 0.52 0.06 0.61 0.52 0.27 0.70 0.33 13:45 0.17 0.07 0.11 0.03 0.19 0.11 0.26 19:45 0.54 0.08 0.62 0.52 0.27 0.71 0.34 13:50 0.18 0.07 0.11 0.03 0.19 0.11 0.27 19:50 0.54 0.10 0.62 0.53 0.28 0.71 0.34 13:55 0.18 0.07 0.11 0.03 0.19 0.11 0.27 19:55 0.54 0.11 0.63 0.53 0.29 0.72 0.35 14:00 0.20 0.07 0.11 0.03 0.18 0.12 0.27 20:00 0.60 0.13 0.63 0.53 0.33 0.74 0.36 14:05 0.19 0.07 0.11 0.04 0.18 0.12 0.27 20:05 0.61 0.13 0.63 0.53 0.33 0.74 0.37 14:10 0.20 0.07 0.11 0.04 0.19 0.13 0.27 20:10 0.63 0.14 0.64 0.56 0.34 0.75 0.39 14:15 0.19 0.07 0.12 0.05 0.18 0.13 0.28 20:15 0.63 0.16 0.65 0.56 0.36 0.77 0.41 14:20 0.18 0.06 0.12 0.04 0.19 0.14 0.28 20:20 0.63 0.17 0.66 0.57 0.37 0.78 0.43 14:25 0.19 0.06 0.13 0.04 0.19 0.14 0.28 20:25 0.64 0.17 0.67 0.57 0.37 0.78 0.43 14:30 0.20 0.08 0.14 0.05 0.19 0.14 0.28 20:30 0.66 0.20 0.67 0.58 0.38 0.80 0.49 14:35 0.19 0.08 0.15 0.05 0.20 0.14 0.28 20:35 0.66 0.20 0.68 0.58 0.39 0.81 0.52 14:40 0.19 0.08 0.15 0.05 0.20 0.14 0.28 20:40 0.67 0.22 0.68 0.58 0.40 0.82 0.52 14:45 0.19 0.09 0.16 0.06 0.20 0.14 0.29 20:45 0.68 0.24 0.70 0.58 0.43 0.85 0.57 14:50 0.21 0.09 0.16 0.06 0.19 0.15 0.30 20:50 0.67 0.24 0.70 0.58 0.43 0.85 0.57 14:55 0.19 0.10 0.17 0.07 0.20 0.14 0.30 20:55 0.68 0.24 0.70 0.59 0.45 0.85 0.58 15:00 0.21 0.09 0.18 0.06 0.19 0.16 0.35 21:00 0.70 0.27 0.72 0.60 0.48 0.87 0.64 15:05 0.21 0.11 0.19 0.05 0.19 0.16 0.35 21:05 0.70 0.28 0.72 0.61 0.48 0.88 0.65 15:10 0.22 0.11 0.20 0.05 0.20 0.16 0.35 21:10 0.70 0.30 0.73 0.63 0.50 0.88 0.66 15:15 0.21 0.11 0.22 0.06 0.20 0.16 0.36 21:15 0.71 0.32 0.74 0.63 0.52 0.89 0.68 15:20 0.24 0.12 0.24 0.06 0.20 0.16 0.40 21:20 0.72 0.34 0.75 0.64 0.53 0.89 0.68 15:25 0.24 0.12 0.24 0.06 0.21 0.16 0.40 21:25 0.72 0.35 0.75 0.64 0.54 0.90 0.69 15:30 0.24 0.11 0.26 0.07 0.21 0.19 0.41 21:30 0.72 0.37 0.76 0.66 0.57 0.91 0.73 15:35 0.25 0.12 0.27 0.07 0.21 0.20 0.41 21:35 0.73 0.38 0.77 0.66 0.58 0.92 0.74 15:40 0.24 0.12 0.28 0.07 0.21 0.21 0.41 21:40 0.74 0.41 0.78 0.67 0.60 0.92 0.75 15:45 0.24 0.11 0.29 0.07 0.22 0.22 0.43 21:45 0.75 0.42 0.79 0.68 0.62 0.93 0.76 15:50 0.23 0.10 0.30 0.07 0.22 0.22 0.43 21:50 0.75 0.43 0.79 0.69 0.64 0.93 0.76 15:55 0.23 0.10 0.30 0.06 0.22 0.23 0.43 21:55 0.75 0.44 0.80 0.70 0.64 0.93 0.77 16:00 0.23 0.08 0.33 0.06 0.22 0.26 0.45 22:00 0.73 0.47 0.81 0.72 0.68 0.95 0.82 16:05 0.24 0.09 0.34 0.06 0.23 0.27 0.46 22:05 0.74 0.48 0.81 0.73 0.69 0.96 0.82 16:10 0.23 0.09 0.35 0.06 0.22 0.28 0.46 22:10 0.76 0.50 0.82 0.74 0.71 0.96 0.82 16:15 0.26 0.11 0.38 0.06 0.22 0.30 0.46 22:15 0.75 0.52 0.83 0.74 0.72 0.96 0.83 16:20 0.26 0.11 0.38 0.06 0.23 0.30 0.46 22:20 0.76 0.54 0.83 0.74 0.74 0.96 0.85 16:25 0.26 0.10 0.39 0.06 0.24 0.31 0.46 22:25 0.77 0.54 0.84 0.75 0.75 0.96 0.85 16:30 0.29 0.11 0.41 0.06 0.24 0.33 0.44 22:30 0.79 0.59 0.85 0.78 0.78 0.97 0.87 16:35 0.28 0.11 0.41 0.06 0.24 0.34 0.45 22:35 0.79 0.62 0.86 0.78 0.79 0.97 0.87 16:40 0.27 0.11 0.42 0.06 0.23 0.35 0.45 22:40 0.80 0.63 0.86 0.77 0.79 0.97 0.88 16:45 0.28 0.11 0.43 0.07 0.23 0.34 0.45 22:45 0.81 0.66 0.87 0.78 0.81 0.97 0.88 16:50 0.28 0.11 0.43 0.07 0.23 0.35 0.45 22:50 0.82 0.66 0.87 0.79 0.82 0.97 0.88 16:55 0.28 0.11 0.43 0.07 0.23 0.34 0.45 22:55 0.83 0.67 0.87 0.79 0.83 0.97 0.89 17:00 0.27 0.08 0.44 0.10 0.22 0.36 0.42 23:00 0.86 0.70 0.89 0.81 0.85 0.98 0.94 17:05 0.27 0.09 0.45 0.10 0.22 0.36 0.42 23:05 0.87 0.72 0.90 0.82 0.86 0.98 0.95 17:10 0.27 0.09 0.46 0.10 0.22 0.37 0.43 23:10 0.88 0.72 0.91 0.82 0.87 0.98 0.95 17:15 0.27 0.09 0.47 0.11 0.21 0.38 0.43 23:15 0.89 0.74 0.91 0.83 0.88 0.98 0.96 17:20 0.28 0.10 0.48 0.12 0.20 0.39 0.43 23:20 0.89 0.76 0.92 0.84 0.89 0.99 0.97 17:25 0.30 0.08 0.48 0.13 0.20 0.39 0.43 23:25 0.89 0.77 0.93 0.85 0.89 0.99 0.97 17:30 0.29 0.09 0.50 0.16 0.18 0.41 0.35 23:30 0.91 0.79 0.94 0.87 0.92 0.99 0.99 17:35 0.29 0.08 0.50 0.17 0.19 0.41 0.34 23:35 0.92 0.80 0.94 0.87 0.92 0.99 0.99 17:40 0.30 0.10 0.51 0.18 0.17 0.42 0.35 23:40 0.93 0.81 0.95 0.88 0.93 0.99 0.99 17:45 0.30 0.10 0.52 0.23 0.19 0.44 0.34 23:45 0.94 0.82 0.95 0.89 0.95 0.99 0.99 17:50 0.30 0.09 0.53 0.23 0.19 0.44 0.35 23:50 0.94 0.83 0.96 0.89 0.95 0.99 0.99 17:55 0.30 0.09 0.53 0.23 0.19 0.45 0.35 23:55 0.95 0.84 0.96 0.89 0.96 0.99 0.99 97 Table A7. Weekend profiles for Night-work Time Under 25 25-34 35-44 45-54 55-64 65-74 Over 75 Time Under 25 25-34 35-44 45-54 55-64 65-74 Over 75 0:00 0.08 0.24 0.11 0.30 0.04 0.02 0.00 6:00 0.87 0.82 0.73 0.72 0.76 0.97 0.98 0:05 0.07 0.23 0.12 0.30 0.04 0.01 0.00 6:05 0.88 0.82 0.74 0.73 0.77 0.97 0.98 0:10 0.08 0.24 0.12 0.30 0.04 0.01 0.00 6:10 0.88 0.83 0.74 0.74 0.78 0.97 0.98 0:15 0.07 0.24 0.13 0.29 0.03 0.01 0.00 6:15 0.89 0.82 0.75 0.74 0.81 0.96 0.98 0:20 0.07 0.23 0.12 0.30 0.03 0.01 0.00 6:20 0.88 0.83 0.74 0.74 0.82 0.96 0.98 0:25 0.07 0.23 0.13 0.30 0.03 0.01 0.00 6:25 0.88 0.83 0.74 0.74 0.82 0.96 0.98 0:30 0.06 0.21 0.13 0.30 0.02 0.00 0.00 6:30 0.88 0.84 0.75 0.75 0.81 0.94 0.97 0:35 0.06 0.22 0.13 0.30 0.02 0.00 0.00 6:35 0.88 0.84 0.75 0.76 0.81 0.94 0.97 0:40 0.07 0.20 0.13 0.30 0.02 0.00 0.00 6:40 0.89 0.84 0.75 0.76 0.81 0.94 0.97 0:45 0.07 0.20 0.13 0.29 0.01 0.00 0.00 6:45 0.89 0.85 0.75 0.76 0.81 0.94 0.97 0:50 0.08 0.20 0.14 0.29 0.01 0.00 0.00 6:50 0.89 0.85 0.75 0.76 0.81 0.94 0.97 0:55 0.08 0.20 0.14 0.29 0.01 0.00 0.00 6:55 0.89 0.85 0.75 0.77 0.81 0.95 0.97 1:00 0.06 0.18 0.16 0.28 0.03 0.00 0.00 7:00 0.88 0.86 0.76 0.77 0.79 0.94 0.97 1:05 0.07 0.18 0.17 0.28 0.04 0.00 0.00 7:05 0.90 0.86 0.76 0.78 0.80 0.91 0.95 1:10 0.08 0.19 0.17 0.28 0.04 0.00 0.00 7:10 0.90 0.87 0.78 0.79 0.79 0.94 0.94 1:15 0.07 0.20 0.16 0.28 0.04 0.01 0.00 7:15 0.91 0.89 0.77 0.82 0.80 0.92 0.94 1:20 0.10 0.22 0.17 0.28 0.04 0.02 0.00 7:20 0.91 0.89 0.77 0.83 0.80 0.92 0.94 1:25 0.10 0.22 0.17 0.28 0.04 0.02 0.00 7:25 0.92 0.90 0.78 0.84 0.82 0.92 0.94 1:30 0.15 0.23 0.18 0.28 0.04 0.03 0.00 7:30 0.93 0.92 0.76 0.85 0.82 0.90 0.92 1:35 0.15 0.23 0.18 0.29 0.06 0.03 0.00 7:35 0.93 0.92 0.77 0.85 0.82 0.90 0.92 1:40 0.15 0.24 0.19 0.29 0.06 0.03 0.00 7:40 0.92 0.92 0.77 0.85 0.84 0.91 0.92 1:45 0.16 0.24 0.19 0.29 0.06 0.03 0.00 7:45 0.92 0.93 0.76 0.85 0.84 0.91 0.92 1:50 0.17 0.24 0.19 0.29 0.06 0.03 0.00 7:50 0.92 0.93 0.77 0.87 0.84 0.91 0.92 1:55 0.16 0.25 0.19 0.29 0.06 0.03 0.00 7:55 0.93 0.93 0.77 0.87 0.84 0.91 0.92 2:00 0.19 0.28 0.22 0.29 0.09 0.04 0.00 8:00 0.91 0.93 0.73 0.87 0.83 0.87 0.89 2:05 0.19 0.30 0.22 0.30 0.10 0.04 0.00 8:05 0.91 0.94 0.73 0.89 0.84 0.88 0.87 2:10 0.19 0.31 0.23 0.30 0.11 0.04 0.00 8:10 0.91 0.94 0.73 0.89 0.84 0.88 0.87 2:15 0.22 0.30 0.23 0.30 0.11 0.05 0.00 8:15 0.90 0.94 0.73 0.89 0.83 0.87 0.85 2:20 0.25 0.32 0.24 0.30 0.11 0.05 0.00 8:20 0.89 0.94 0.73 0.90 0.84 0.88 0.85 2:25 0.24 0.31 0.24 0.30 0.11 0.05 0.00 8:25 0.88 0.94 0.73 0.90 0.84 0.88 0.85 2:30 0.25 0.33 0.25 0.31 0.11 0.04 0.02 8:30 0.87 0.94 0.71 0.87 0.86 0.83 0.87 2:35 0.26 0.33 0.25 0.31 0.11 0.05 0.02 8:35 0.87 0.93 0.71 0.88 0.86 0.83 0.87 2:40 0.27 0.34 0.25 0.31 0.12 0.05 0.02 8:40 0.86 0.94 0.72 0.88 0.86 0.83 0.87 2:45 0.29 0.35 0.26 0.32 0.12 0.06 0.02 8:45 0.87 0.94 0.72 0.88 0.85 0.84 0.84 2:50 0.30 0.35 0.27 0.32 0.12 0.05 0.02 8:50 0.85 0.94 0.70 0.88 0.84 0.83 0.82 2:55 0.30 0.35 0.27 0.32 0.12 0.05 0.02 8:55 0.85 0.94 0.71 0.88 0.84 0.83 0.82 3:00 0.33 0.36 0.27 0.34 0.17 0.05 0.03 9:00 0.84 0.90 0.67 0.84 0.79 0.81 0.82 3:05 0.33 0.36 0.27 0.34 0.17 0.05 0.03 9:05 0.84 0.90 0.66 0.84 0.78 0.81 0.82 3:10 0.33 0.37 0.28 0.35 0.18 0.05 0.03 9:10 0.83 0.89 0.67 0.84 0.79 0.81 0.81 3:15 0.34 0.37 0.28 0.35 0.18 0.05 0.03 9:15 0.84 0.89 0.66 0.84 0.79 0.76 0.81 3:20 0.34 0.38 0.28 0.36 0.18 0.05 0.03 9:20 0.84 0.88 0.65 0.84 0.77 0.77 0.79 3:25 0.34 0.39 0.28 0.36 0.18 0.05 0.03 9:25 0.83 0.88 0.65 0.84 0.78 0.77 0.81 3:30 0.36 0.39 0.28 0.36 0.18 0.05 0.05 9:30 0.83 0.87 0.61 0.82 0.76 0.72 0.81 3:35 0.37 0.40 0.28 0.36 0.19 0.06 0.06 9:35 0.82 0.87 0.61 0.81 0.76 0.71 0.82 3:40 0.38 0.40 0.28 0.37 0.19 0.06 0.06 9:40 0.82 0.87 0.61 0.82 0.76 0.72 0.82 3:45 0.39 0.40 0.29 0.37 0.19 0.08 0.06 9:45 0.81 0.87 0.61 0.81 0.76 0.71 0.77 3:50 0.40 0.41 0.29 0.37 0.19 0.10 0.06 9:50 0.81 0.87 0.61 0.81 0.76 0.71 0.79 3:55 0.40 0.42 0.29 0.37 0.19 0.10 0.06 9:55 0.81 0.86 0.61 0.81 0.76 0.71 0.79 4:00 0.82 0.79 0.80 0.70 0.77 0.95 0.98 10:00 0.74 0.85 0.58 0.79 0.76 0.70 0.73 4:05 0.82 0.79 0.80 0.70 0.77 0.95 0.98 10:05 0.74 0.85 0.58 0.78 0.76 0.70 0.71 4:10 0.82 0.79 0.80 0.70 0.77 0.95 0.98 10:10 0.76 0.86 0.59 0.79 0.76 0.71 0.71 4:15 0.83 0.79 0.81 0.71 0.77 0.95 0.98 10:15 0.76 0.86 0.60 0.79 0.74 0.70 0.71 4:20 0.84 0.79 0.81 0.71 0.77 0.95 0.98 10:20 0.76 0.86 0.59 0.79 0.76 0.71 0.71 4:25 0.84 0.79 0.81 0.71 0.77 0.95 0.98 10:25 0.76 0.86 0.59 0.79 0.76 0.71 0.71 4:30 0.84 0.79 0.79 0.71 0.78 0.95 1.00 10:30 0.74 0.85 0.55 0.78 0.75 0.68 0.68 4:35 0.84 0.79 0.79 0.71 0.78 0.95 1.00 10:35 0.75 0.86 0.56 0.78 0.74 0.69 0.69 4:40 0.84 0.79 0.78 0.71 0.78 0.95 1.00 10:40 0.74 0.86 0.56 0.78 0.76 0.69 0.71 4:45 0.85 0.79 0.78 0.71 0.78 0.95 1.00 10:45 0.74 0.86 0.53 0.77 0.76 0.69 0.71 4:50 0.85 0.79 0.78 0.71 0.78 0.96 1.00 10:50 0.74 0.86 0.54 0.77 0.76 0.71 0.71 4:55 0.85 0.79 0.78 0.71 0.78 0.96 1.00 10:55 0.74 0.86 0.54 0.77 0.76 0.70 0.71 5:00 0.86 0.79 0.76 0.71 0.78 0.96 1.00 11:00 0.69 0.82 0.54 0.76 0.75 0.69 0.71 5:05 0.86 0.80 0.76 0.72 0.79 0.96 1.00 11:05 0.70 0.83 0.54 0.76 0.75 0.71 0.71 5:10 0.86 0.80 0.76 0.72 0.79 0.96 1.00 11:10 0.69 0.84 0.54 0.76 0.75 0.71 0.71 5:15 0.86 0.80 0.76 0.72 0.78 0.96 1.00 11:15 0.68 0.83 0.53 0.75 0.75 0.71 0.73 5:20 0.86 0.80 0.77 0.72 0.78 0.96 1.00 11:20 0.68 0.81 0.53 0.74 0.74 0.70 0.74 5:25 0.87 0.80 0.77 0.72 0.78 0.96 1.00 11:25 0.67 0.81 0.54 0.74 0.74 0.70 0.74 5:30 0.87 0.81 0.75 0.72 0.76 0.96 1.00 11:30 0.65 0.78 0.53 0.74 0.73 0.69 0.73 5:35 0.87 0.81 0.75 0.72 0.77 0.96 1.00 11:35 0.66 0.78 0.53 0.74 0.74 0.69 0.71 5:40 0.87 0.81 0.75 0.72 0.77 0.96 1.00 11:40 0.67 0.79 0.54 0.75 0.74 0.69 0.71 5:45 0.87 0.81 0.74 0.72 0.77 0.97 1.00 11:45 0.66 0.79 0.55 0.76 0.75 0.69 0.71 5:50 0.87 0.81 0.74 0.72 0.77 0.97 1.00 11:50 0.67 0.78 0.56 0.77 0.74 0.68 0.73 5:55 0.87 0.81 0.74 0.72 0.77 0.97 1.00 11:55 0.68 0.78 0.56 0.77 0.74 0.67 0.73 98 Table A7(cont’d) Time Under 25 25-34 35-44 45-54 55-64 65-74 Over 75 Time Under 25 25-34 35-44 45-54 55-64 65-74 Over 75 12:00 0.65 0.72 0.54 0.77 0.72 0.65 0.73 18:00 0.56 0.61 0.52 0.70 0.60 0.69 0.71 12:05 0.65 0.72 0.55 0.78 0.72 0.65 0.73 18:05 0.56 0.63 0.50 0.69 0.60 0.68 0.71 12:10 0.66 0.73 0.56 0.78 0.73 0.66 0.73 18:10 0.58 0.62 0.50 0.69 0.59 0.69 0.71 12:15 0.65 0.72 0.56 0.78 0.72 0.67 0.73 18:15 0.58 0.62 0.49 0.70 0.60 0.69 0.71 12:20 0.65 0.72 0.57 0.79 0.72 0.68 0.74 18:20 0.60 0.63 0.49 0.70 0.59 0.67 0.71 12:25 0.65 0.72 0.57 0.79 0.72 0.69 0.76 18:25 0.59 0.63 0.50 0.70 0.60 0.67 0.71 12:30 0.63 0.70 0.56 0.78 0.69 0.67 0.77 18:30 0.59 0.60 0.45 0.68 0.58 0.60 0.68 12:35 0.63 0.70 0.58 0.79 0.68 0.67 0.81 18:35 0.58 0.60 0.45 0.68 0.58 0.61 0.68 12:40 0.63 0.71 0.58 0.78 0.69 0.68 0.81 18:40 0.58 0.60 0.44 0.68 0.57 0.62 0.68 12:45 0.62 0.71 0.58 0.78 0.69 0.68 0.82 18:45 0.59 0.59 0.44 0.67 0.56 0.63 0.69 12:50 0.62 0.72 0.57 0.78 0.71 0.67 0.82 18:50 0.59 0.60 0.43 0.68 0.55 0.62 0.71 12:55 0.62 0.72 0.58 0.78 0.71 0.69 0.82 18:55 0.58 0.60 0.44 0.69 0.54 0.62 0.73 13:00 0.59 0.71 0.55 0.76 0.71 0.67 0.81 19:00 0.55 0.57 0.41 0.63 0.53 0.56 0.68 13:05 0.59 0.71 0.56 0.78 0.70 0.68 0.76 19:05 0.55 0.58 0.42 0.63 0.53 0.56 0.68 13:10 0.58 0.72 0.56 0.77 0.70 0.69 0.76 19:10 0.56 0.58 0.42 0.63 0.52 0.55 0.68 13:15 0.59 0.72 0.56 0.78 0.70 0.70 0.76 19:15 0.58 0.58 0.42 0.60 0.52 0.56 0.66 13:20 0.60 0.72 0.57 0.78 0.70 0.70 0.77 19:20 0.58 0.58 0.42 0.60 0.52 0.56 0.66 13:25 0.60 0.72 0.57 0.78 0.71 0.70 0.77 19:25 0.57 0.59 0.41 0.60 0.52 0.57 0.66 13:30 0.60 0.71 0.56 0.78 0.67 0.68 0.79 19:30 0.57 0.59 0.42 0.58 0.49 0.55 0.60 13:35 0.60 0.72 0.56 0.79 0.67 0.68 0.81 19:35 0.57 0.59 0.41 0.57 0.49 0.54 0.60 13:40 0.61 0.73 0.56 0.78 0.69 0.69 0.82 19:40 0.56 0.59 0.42 0.56 0.49 0.51 0.58 13:45 0.63 0.73 0.56 0.78 0.69 0.69 0.82 19:45 0.56 0.58 0.42 0.56 0.47 0.52 0.58 13:50 0.64 0.73 0.57 0.78 0.70 0.70 0.84 19:50 0.56 0.58 0.42 0.56 0.46 0.52 0.58 13:55 0.64 0.73 0.57 0.77 0.70 0.70 0.85 19:55 0.58 0.58 0.43 0.55 0.46 0.54 0.58 14:00 0.60 0.72 0.58 0.76 0.67 0.70 0.84 20:00 0.51 0.54 0.42 0.51 0.42 0.48 0.48 14:05 0.59 0.72 0.58 0.77 0.68 0.70 0.85 20:05 0.51 0.55 0.42 0.51 0.42 0.49 0.48 14:10 0.61 0.74 0.58 0.78 0.66 0.70 0.85 20:10 0.50 0.54 0.42 0.50 0.41 0.49 0.48 14:15 0.62 0.75 0.59 0.78 0.66 0.70 0.84 20:15 0.48 0.54 0.41 0.49 0.39 0.49 0.50 14:20 0.61 0.76 0.59 0.79 0.66 0.70 0.85 20:20 0.48 0.55 0.40 0.49 0.38 0.49 0.50 14:25 0.61 0.76 0.59 0.81 0.66 0.70 0.84 20:25 0.48 0.54 0.40 0.48 0.38 0.49 0.48 14:30 0.61 0.73 0.60 0.80 0.64 0.71 0.84 20:30 0.47 0.49 0.38 0.44 0.36 0.46 0.47 14:35 0.63 0.74 0.60 0.80 0.66 0.71 0.84 20:35 0.45 0.49 0.38 0.45 0.36 0.46 0.47 14:40 0.63 0.74 0.60 0.80 0.66 0.71 0.84 20:40 0.44 0.49 0.36 0.44 0.36 0.48 0.47 14:45 0.63 0.74 0.58 0.80 0.67 0.70 0.84 20:45 0.44 0.48 0.35 0.43 0.36 0.44 0.47 14:50 0.64 0.74 0.57 0.80 0.67 0.70 0.84 20:50 0.44 0.48 0.34 0.43 0.36 0.44 0.47 14:55 0.63 0.74 0.57 0.80 0.67 0.71 0.84 20:55 0.44 0.48 0.34 0.42 0.36 0.43 0.44 15:00 0.61 0.72 0.56 0.77 0.64 0.70 0.81 21:00 0.39 0.40 0.27 0.35 0.29 0.38 0.35 15:05 0.62 0.72 0.57 0.77 0.65 0.70 0.81 21:05 0.40 0.40 0.28 0.34 0.29 0.38 0.34 15:10 0.62 0.73 0.57 0.78 0.64 0.71 0.81 21:10 0.38 0.40 0.27 0.34 0.28 0.35 0.32 15:15 0.61 0.73 0.58 0.78 0.64 0.72 0.81 21:15 0.36 0.39 0.27 0.33 0.28 0.33 0.31 15:20 0.61 0.72 0.58 0.78 0.66 0.74 0.81 21:20 0.35 0.38 0.27 0.32 0.28 0.33 0.29 15:25 0.61 0.72 0.58 0.78 0.66 0.74 0.81 21:25 0.35 0.38 0.27 0.32 0.28 0.32 0.29 15:30 0.58 0.70 0.59 0.78 0.65 0.73 0.81 21:30 0.31 0.33 0.24 0.30 0.25 0.27 0.26 15:35 0.59 0.71 0.59 0.78 0.66 0.73 0.81 21:35 0.31 0.33 0.23 0.29 0.25 0.27 0.24 15:40 0.58 0.70 0.60 0.79 0.67 0.72 0.81 21:40 0.31 0.33 0.23 0.26 0.23 0.25 0.24 15:45 0.58 0.69 0.60 0.78 0.67 0.72 0.81 21:45 0.31 0.32 0.23 0.27 0.22 0.25 0.24 15:50 0.59 0.70 0.60 0.79 0.67 0.73 0.81 21:50 0.30 0.32 0.22 0.28 0.21 0.24 0.23 15:55 0.59 0.71 0.60 0.78 0.66 0.75 0.84 21:55 0.30 0.32 0.22 0.28 0.20 0.24 0.23 16:00 0.58 0.68 0.58 0.77 0.66 0.73 0.82 22:00 0.24 0.22 0.19 0.23 0.16 0.20 0.11 16:05 0.60 0.68 0.59 0.78 0.66 0.73 0.82 22:05 0.24 0.23 0.19 0.24 0.16 0.19 0.11 16:10 0.58 0.70 0.60 0.78 0.67 0.73 0.81 22:10 0.24 0.23 0.19 0.24 0.16 0.18 0.11 16:15 0.54 0.69 0.60 0.79 0.67 0.72 0.84 22:15 0.24 0.23 0.18 0.25 0.14 0.14 0.10 16:20 0.55 0.69 0.60 0.79 0.67 0.72 0.84 22:20 0.23 0.23 0.18 0.25 0.13 0.14 0.10 16:25 0.56 0.70 0.61 0.79 0.68 0.72 0.84 22:25 0.23 0.22 0.17 0.26 0.13 0.14 0.10 16:30 0.55 0.68 0.58 0.78 0.65 0.72 0.81 22:30 0.19 0.21 0.15 0.24 0.11 0.10 0.08 16:35 0.57 0.68 0.58 0.78 0.65 0.71 0.81 22:35 0.19 0.21 0.14 0.24 0.12 0.09 0.08 16:40 0.57 0.70 0.58 0.78 0.66 0.71 0.81 22:40 0.20 0.20 0.14 0.25 0.12 0.10 0.06 16:45 0.56 0.69 0.58 0.78 0.66 0.71 0.84 22:45 0.20 0.20 0.14 0.24 0.12 0.11 0.06 16:50 0.55 0.68 0.60 0.78 0.66 0.72 0.85 22:50 0.20 0.21 0.14 0.24 0.12 0.10 0.06 16:55 0.56 0.69 0.59 0.77 0.66 0.74 0.84 22:55 0.20 0.21 0.14 0.24 0.12 0.09 0.06 17:00 0.55 0.65 0.57 0.75 0.64 0.70 0.82 23:00 0.18 0.20 0.13 0.25 0.10 0.05 0.02 17:05 0.55 0.64 0.57 0.75 0.63 0.70 0.82 23:05 0.18 0.21 0.13 0.26 0.09 0.04 0.02 17:10 0.54 0.65 0.57 0.75 0.64 0.71 0.81 23:10 0.18 0.20 0.13 0.26 0.08 0.04 0.02 17:15 0.55 0.65 0.57 0.75 0.64 0.70 0.79 23:15 0.19 0.21 0.14 0.26 0.07 0.03 0.00 17:20 0.56 0.66 0.57 0.75 0.64 0.71 0.79 23:20 0.18 0.21 0.13 0.27 0.07 0.02 0.00 17:25 0.57 0.66 0.57 0.76 0.64 0.71 0.79 23:25 0.17 0.21 0.13 0.28 0.06 0.02 0.00 17:30 0.55 0.66 0.53 0.75 0.59 0.70 0.71 23:30 0.16 0.22 0.12 0.28 0.07 0.02 0.00 17:35 0.55 0.65 0.54 0.74 0.59 0.70 0.71 23:35 0.18 0.22 0.12 0.28 0.07 0.02 0.00 17:40 0.56 0.63 0.55 0.76 0.61 0.70 0.71 23:40 0.18 0.23 0.11 0.29 0.07 0.02 0.00 17:45 0.55 0.62 0.55 0.74 0.59 0.70 0.71 23:45 0.18 0.24 0.11 0.31 0.08 0.02 0.00 17:50 0.56 0.62 0.55 0.74 0.60 0.69 0.71 23:50 0.18 0.26 0.12 0.31 0.08 0.02 0.00 17:55 0.56 0.63 0.54 0.75 0.59 0.71 0.71 23:55 0.19 0.26 0.13 0.32 0.08 0.02 0.00 99 Number of times leaving Number of times leaving Number of times leaving 1.0 1 1.0 1 1.0 1 0.9 0.9 0.9 0.8 0.8 0.8 0.8 0.8 0.8 0.7 0.7 0.7 0.6 0.6 0.6 0.6 0.6 0.6 0.5 0.5 0.5 0.4 0.4 0.4 0.4 0.4 0.4 0.3 0.3 0.3 0.2 0.2 0.2 0.2 0.2 0.2 0.1 0.1 0.1 0.0 0 0 1 2 3 4 5 6 7 8 9 10 0.0 0 0.0 0 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 (a) (b) (c) Number of times leaving 1.0 Number of times leaving 1 Number of times leaving 1.0 1 1.0 1 0.9 0.9 0.9 0.8 0.8 0.80.8 0.8 0.8 0.7 0.7 0.7 0.6 0.6 0.60.6 0.6 0.6 0.5 0.5 0.5 0.4 0.4 0.40.4 0.4 0.4 0.3 0.3 0.3 0.2 0.2 0.20.2 0.2 0.2 0.1 0.1 0.1 0.0 0 0.0 0 0.0 0 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 (d) (e) (f) Figure B1. Number of Departures from home for people of age group (a) 25 to 34, (b) 35 to 44, (c) 45 to 54, (d) 55 to 64, (e) 65 to 74 and (f) over 75 years for weekday Stay-home profiles Number of times leaving Number of times leaving 1 Number of times leaving 1.0 1 1.0 1.0 1 0.9 0.9 0.9 0.8 0.8 0.8 0.80.8 0.8 0.7 0.7 0.7 0.60.6 0.6 0.6 0.6 0.6 0.5 0.5 0.5 0.40.4 0.4 0.4 0.4 0.4 0.3 0.3 0.3 0.20.2 0.2 0.2 0.2 0.2 0.1 0.1 0.1 0.0 0 0.0 0 0 1 2 3 4 5 6 7 8 9 10 0.0 0 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 (a) (b) (c) Number of times leaving Number of times leaving Number of times leaving 1.0 1.0 1 1.0 1 1 0.9 0.9 0.9 0.80.8 0.80.8 0.8 0.8 0.7 0.7 0.7 0.60.6 0.60.6 0.6 0.6 0.5 0.5 0.5 0.40.4 0.40.4 0.4 0.4 0.3 0.3 0.3 0.20.2 0.20.2 0.2 0.2 0.1 0.1 0.1 0.0 0 0.0 0 0 1 2 3 4 5 6 7 8 9 10 0.0 0 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 (d) (e) (f) Figure B2. Number of Departures from home for people of age group (a) 25 to 34, (b) 35 to 44, (c) 45 to 54, (d) 55 to 64, (e) 65 to 74 and (f) over 75 years for weekdays Day-work profiles 100 Number of times leaving Number of times leaving Number of times leaving 1.0 1 1 1 1.0 1.0 0.9 0.9 0.9 0.80.8 0.8 0.8 0.8 0.8 0.7 0.7 0.7 0.60.6 0.6 0.6 0.6 0.6 0.5 0.5 0.5 0.40.4 0.4 0.4 0.4 0.4 0.3 0.3 0.3 0.20.2 0.2 0.2 0.2 0.2 0.1 0.1 0.1 0.0 0 0 1 2 3 4 5 6 7 8 9 10 0.0 0 0 1 2 3 4 5 6 7 8 9 10 0.0 0 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 (a) (b) (c) Number of times leaving Number of times leaving Number of times leaving 1.0 1.0 1.0 1 1 1 0.9 0.9 0.9 0.8 0.8 0.8 0.8 0.8 0.8 0.7 0.7 0.7 0.6 0.6 0.6 0.6 0.6 0.6 0.5 0.5 0.5 0.4 0.4 0.4 0.4 0.4 0.4 0.3 0.3 0.3 0.2 0.2 0.2 0.2 0.2 0.2 0.1 0.1 0.1 0.0 0 0 1 2 3 4 5 6 7 8 9 10 0.0 0 0 1 2 3 4 5 6 7 8 9 10 0.0 0 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 (d) (e) (f) Figure B3. Number of Departures from home for people of age group (a) 25 to 34, (b) 35 to 44, (c) 45 to 54, (d) 55 to 64, (e) 65 to 74 and (f) over 75 years for weekday Night-work profiles Number of times leaving Number of times leaving Number of times leaving 1.0 1 1 1 1.0 1.0 0.9 0.9 0.9 0.8 0.8 0.8 0.8 0.8 0.8 0.7 0.7 0.7 0.6 0.6 0.6 0.6 0.6 0.6 0.5 0.5 0.5 0.4 0.4 0.4 0.4 0.4 0.4 0.3 0.3 0.3 0.2 0.2 0.2 0.2 0.2 0.2 0.1 0.1 0.1 0.0 0 0 1 2 3 4 5 6 7 8 9 10 0.0 0 0 1 2 3 4 5 6 7 8 9 10 0.0 0 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 (a) (b) (c) Number of times leaving Number of times leaving Number of times leaving 1.0 1.0 1.0 1 1 1 0.9 0.9 0.9 0.8 0.8 0.8 0.8 0.8 0.8 0.7 0.7 0.7 0.6 0.6 0.6 0.6 0.6 0.6 0.5 0.5 0.5 0.4 0.4 0.4 0.4 0.4 0.4 0.3 0.3 0.3 0.2 0.2 0.2 0.2 0.2 0.2 0.1 0.1 0.1 0.0 0 0 1 2 3 4 5 6 7 8 9 10 0.0 0 0.0 0 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 (d) (e) (f) Figure B4. Number of Departures from home for people of age group (a) 25 to 34, (b) 35 to 44, (c) 45 to 54, (d) 55 to 64, (e) 65 to 74 and (f) over 75 years for weekend Stay-home profiles 101 Number of times leaving Number of times leaving Number of times leaving 1.0 1 1.0 1 1.0 1 0.9 0.9 0.9 0.8 0.8 0.8 0.8 0.8 0.8 0.7 0.7 0.7 0.6 0.6 0.6 0.6 0.6 0.6 0.5 0.5 0.5 0.4 0.4 0.4 0.4 0.4 0.4 0.3 0.3 0.3 0.2 0.2 0.2 0.2 0.2 0.2 0.1 0.1 0.1 0.0 0 0.0 0 0.0 0 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 (a) Number of times leaving (b) Number of times leaving (c) Number of times leaving 1.0 1 1.0 1 1.0 1 0.9 0.9 0.9 0.8 0.8 0.8 0.8 0.8 0.8 0.7 0.7 0.7 0.6 0.6 0.6 0.6 0.6 0.6 0.5 0.5 0.5 0.4 0.4 0.4 0.4 0.4 0.4 0.3 0.3 0.3 0.2 0.2 0.2 0.2 0.2 0.2 0.1 0.1 0.1 0.0 0 0.0 0 0.0 0 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 (d) (e) (f) Figure B5. Number of Departures from home for people of age group (a) 25 to 34, (b) 35 to 44, (c) 45 to 54, (d) 55 to 64, (e) 65 to 74 and (f) over 75 years for weekend Day-work profiles Number of times leaving Number of times leaving Number of times leaving 1.0 1.0 1.0 1 1 1 0.9 0.9 0.9 0.8 0.8 0.8 0.8 0.8 0.8 0.7 0.7 0.7 0.6 0.6 0.6 0.6 0.6 0.6 0.5 0.5 0.5 0.4 0.4 0.4 0.4 0.4 0.4 0.3 0.3 0.3 0.2 0.2 0.2 0.2 0.2 0.2 0.1 0.1 0.1 0.0 0 0 1 2 3 4 5 6 7 8 9 10 0.0 0 0 1 2 3 4 5 6 7 8 9 10 0.0 0 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 (a) (b) (c) Number of times leaving Number of times leaving Number of times leaving 1.0 1.0 1.0 1 1 1 0.9 0.9 0.9 0.80.8 0.8 0.8 0.8 0.8 0.7 0.7 0.7 0.60.6 0.6 0.6 0.6 0.6 0.5 0.5 0.5 0.40.4 0.4 0.4 0.4 0.4 0.3 0.3 0.3 0.20.2 0.2 0.2 0.2 0.2 0.1 0.1 0.1 0.0 0 0 1 2 3 4 5 6 7 8 9 10 0.0 0 0 1 2 3 4 5 6 7 8 9 10 0.0 0 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 (d) (e) (f) Figure B6. Number of Departures from home for people of age group (a) 25 to 34, (b) 35 to 44, (c) 45 to 54, (d) 55 to 64, (e) 65 to 74 and (f) over 75 years for weekend Night-work profiles 102 4. CHAPTER 4 - VARIATION IN RESIDENTIAL OCCUPANCY PROFILES IN THE UNITED STATES BY HOUSEHOLD INCOME LEVEL AND CHARACTERISTICS Debrudra Mitra, Yiyi Chu, Kristen Cetin, Yu Wang & Chien-fei Chen (2021) Variation in residential occupancy profiles in the United States by household income level and characteristics, Journal of Building Performance Simulation, 14:6, 692- 711, DOI: 10.1080/19401493.2021.2001572 1. Abstract Accurate representation of occupancy schedules is needed to assess potential energy savings from occupancy-based controls in residential buildings. In this study the variation in U.S. residential occupancy profiles is developed by household income level, age group, household size, and day of the week using 14 years of American Time Use Survey and Current Population Survey data. Based on cluster analysis results, the most common occupancy profiles were Day absence and Stay home where, the time of absence varies from less than 5 hours to 15 hours per day. Low-income individuals and households spent significantly more time at home compared to higher income groups. Finally, a preliminary survey conducted to analyze the impacts of the COVID-19 pandemic suggests that it has substantially impacted occupancy patterns and is likely to do so post- pandemic as well. The results of this research help improve the representation of occupancy schedules in building energy simulation methods. 2. Introduction The building sector is one of the major consumers of energy throughout the world. In the United States, the percentage of energy consumed by the building sector has increased significantly as compared to the other sectors in the past 70 years [1]. The 2020 U.S. Annual Energy Outlook reported that electricity used by residential and commercial buildings is projected to increase at an average of 0.6% and 0.8% per year until 2050 [2]. In addition, energy consumption from the building sector accounts for approximately 30% of all greenhouse gas (GHG) emissions worldwide [3]. In 2018, 12.3% of the overall GHG emissions from direct energy consumed originated from the building sector in the U.S. [4]. Given the need to curb GHG emissions to 103 reduce the impacts of climate change, it is important to develop and implement strategies to reduce the energy consumption of the U.S. building stock [5]. Residential buildings consume more than half of the total energy used by buildings in the United States [2], thus they represent a strong target for improvements to energy efficiency. Overall, among the end uses in these buildings, the heating, ventilation and air conditioning (HVAC) systems consume the most, on average [6]. However, the amount of HVAC energy consumed depends on occupants, their thermostat preferences, as well as how they utilize the building and its energy-consuming systems [7]. In the IEA EBC Annex 53 (Total energy use in buildings – analysis and evaluation methods) report, six key parameters were identified as influencing building energy consumption. These include the (a) building envelope, (b) building equipment, (c) operation and maintenance, (d) indoor comfort criteria, (e) occupant behavior and (f) weather conditions [8]. Three of these six factors: (e) occupant behavior, (d) indoor comfort criteria and (c) operation and maintenance were identified as having a stronger impact on building energy performance as compared to other factors [9, 10]. This suggests that occupants are of significant importance to building energy consumption, and that detection of occupants and adjustment of building system controls based on this knowledge provides significant opportunities for energy savings. For example, Lo et al. (2010) found that accurate occupancy detection in open office spaces can reduce the total energy consumption by up to 30% [11]. For residential applications, Li et al. (2007) studied 25 similar residential buildings in Beijing, China, finding that due to how occupants interact with their HVAC system and setpoints, electricity consumption varied from 0 to 14 kWh/m2 [12]. Other studies show a similar influence of occupants on residential building HVAC utilization profiles [13, 14]. When comparing residential and commercial buildings, several studies have shown that the impact of occupant behavior on the energy consumption in residential buildings is more significant [15-17]. Such findings demonstrate the importance of an accurate understanding and representation of occupancy scenarios, to support evaluating the energy performance of residential buildings. While residential buildings and their occupancy profiles are often assumed to follow “typical” occupancy schedules, a growing body of research suggests that occupant behaviors and schedules can vary substantially across different socio-economic population groups [18]. This is particularly important for low-income households (LIHs), who are more likely to be impacted by the built 104 environment, energy used and the resulting energy bills [19, 20]. Across United States, the median household energy burden, i.e., the percentage of household income that is spent on energy bills, is approximately 2.3% [21]. For LIHs this can be as high as 8.1% [21]. LIHs are also more likely to live in less efficient housing with older and less energy-efficient appliances than higher income households (HIHs) [22]. This emphasizes the importance of understanding variations in occupancy profiles in residential buildings, specifically for LIHs. Currently, most building energy modeling software programs use existing occupancy profiles for energy modeling of residential buildings based on the U.S. Department of Energy’s Reference Building Models [23] and Prototype Building Models [24]. These occupancy profiles for residential buildings originated from the 1989 version of ASHRAE 90.1 [25] and the more recent ASHRAE Advanced Energy Design Guides (AEDG) studies [26]. Another energy simulation tool, BEopt [27, 28], uses occupancy schedules provided in the Building America Housing Simulation Protocol [24]. Generally, these schedules provide an occupancy fraction on an hourly basis; this represents the fraction of total occupants present in the space. This occupancy fraction value ranges from 0 to 1, where 1 represents when all occupants are present at home and 0 represents when no one is present. However, there is limited information discussing the origin of the currently utilized residential occupancy schedules, other than that the schedules were developed using engineering judgement and generalized assumptions based on other data sources [29]. Given that occupant behavior variations in residential building are suggested to lead to up to 30% differences between model-predicated and actual energy use [30], further studies are needed to evaluate the occupancy profile variations in residential buildings. Recently, studies have used probabilistic, data-driven, and machine learning methods [31] for occupancy schedule development in residential buildings, using sensor-based data, and/or by interviewing occupants. Markov models are among the most common methodologies to develop stochastic occupancy schedules. A simple Markov chain for a single occupant was developed by Page et al. [32], which was updated for multi-occupant scenarios by Richardson et al. [33]. Widen et al. [34] developed a stochastic model using non-homogeneous time dependent transition probabilities, which was expanded upon by Lopez-Rodriguez et al. [35] using time use data from Spain [36]. Other variations of Markov chain models, including dynamic Markov models and semi-Markov models, have been implemented to predict occupancy scenarios [37,38]. Similarly, 105 data-driven methodologies such as cluster analysis [39-44], neural networks [40-42], Support Vector Machine (SVM), K-Nearest Neighbor, and C4.5 Decision Tree algorithms [43,46-48] have also been used. However, while these methods are useful in modeling occupant behavior, these studies generally have focused on specific occupied buildings and their occupancy patterns, based on field-collected data, rather than a representative sample of buildings. Using specific building data, it is thus challenging to map such results to the overall population of the United States [49]. Such mapping is desirable for use in representing occupancy patterns of the U.S. population, and subsets thereof. In addition, there are limited current studies that assess overall variations in occupancy patterns across different income groups. To better represent occupancy schedules, the dataset utilized needs to be statistically designed to represent different income groups’ profiles. As such, the main objective of this study is to evaluate the occupancy profile characteristics of LIHs in the U.S., as compared to those considered to be typical, and HIHs using several national representative datasets. This research focuses on defining the following across income groups: how occupancy patterns vary on weekdays and weekends; and how age and household size impact occupancy profiles. The American Time Use Survey (ATUS) and Current Population Survey (CPS) data from 2006 to 2019 were utilized to classify occupancy profiles across income groups for various age groups using cluster analysis methods. Residential Energy Consumption Survey (RECS) data was also used for occupant demographic distributions in residential buildings. However, the COVID-19 pandemic has significantly impacted people’s schedules, and thus the occupancy of their home. As mentioned, [50], approximately 71% of the workforce were working from home in the end of 2020. Another study found that 94% of the global office workers were working from home during the pandemic [51]. Other studies have reported that this change in the work culture improves their performance [52], and that people also prefer to have flexible work schedule [53-57]. Due to this, permanent changes in employer expectations of in-person versus remote work and schedules have occurred and continue to occur, which will impact the occupancy of residential buildings moving forward, beyond the pandemic. 106 The paper is organized as follows. First the datasets utilized, and income definitions are described. Next the cluster analysis methodology is described, followed by the occupancy profile results, including a comparison among occupancy profiles for different income groups, and a more detailed analysis of different occupancy schedule patterns. Next, the most common household members and corresponding household occupancy schedules are discussed. This is followed by initial considerations of the impact of the COVID-19 on occupancy scenarios in residential buildings, based on survey data collected, as this is likely to have long-lasting implications on occupancy schedules moving forward. 3. Datasets 3.1. American Time Use Survey (ATUS) and Current Population Survey (CPS) Conducted by the U.S. Census Bureau, both the CPS and the ATUS began data collection in 2003 [58, 59] and are ongoing every year. The CPS gathers information about the U.S. population, including employment, income, education, and other economic and social well-being data. This dataset includes demographic information, and household income. From the CPS survey, households that have completed 8 months of the survey are eligible to participate in the ATUS. The objective of the ATUS is to collect information on how the U.S. population spends their time. The data are collected each year through a combination of email, telephone and in-person interviews where participants are asked to document what activities they did over a 24-hour period (4:00 am to 3:59 am the following day), and when and where these activities occurred. Categories include 470 types of activities, such as sleeping, housework, and childcare. Demographic data including age and gender are also collected, however, those under the age of 15 or over 75 are designated as 15 and 75 respectively, in the data. The advantage of both the CPS and ATUS datasets is that they statistically represent the overall U.S. population. We note, however, the method of statistical weightage parameter calculation was modified slightly in 2006, thus only 2006 to 2019 data was used in this study. The ATUS data can be mapped to CPS data using common identifier in both datasets, thus the reported household income from the CPS can be used to classify households as being low- or high-income, based on the number of household members [60]. 107 The maximum income for being considered a LIH (i.e., the poverty threshold, based on U.S. government defined values) varied slightly each year, the range of which is shown in Table 1, between 2006 and 2019. As the poverty levels for Hawaii and Alaska differed from the other 48 states, they are shown separately. However, the percentage of participants in Hawaii and Alaska were comparatively lower, thus the data for the remaining 48 states was used in this study to identify the LIHs. Overall, income ranges for HIHs were defined based on household designations obtained from previous studies [61, 62] also shown in Table 1. We note that LIH thresholds vary from year to year, but the HIH thresholds do not; this is because there is no officially defined threshold for HIH, thus we have used a set value, as described in [61,62]. Table 4-1. Household income level (U.S. dollars) for LIH and HIHs between 2006-2019 (Note: MIH income levels are between the LIH and HIH threshold values shown) Maximum Threshold for LIH # of members Alaska Hawaii Other 48 states Minimum Threshold for HIH 1 person 12,250-15,600 11,270-14,380 9,800-12,140 67,500 2 persons 16,500-21,130 15,180-19,460 13,200-16,910 87,500 3 persons 20,750-26,660 19,090-24,450 16,600-21,130 125,000 4 persons 25,000-32,190 23,000-29,620 20,000-25,750 150,000 5 persons 29,250-37,720 26,910-34,700 23,400-30,170 150,000 3.2. Residential Energy Consumption Survey (RECS) The RECS dataset is collected and compiled approximately every six years, starting in 1978, collecting information on residential buildings throughout the United States. This unique dataset provides, through survey data collection, an understanding of building characteristics, building equipment and appliances, occupant behaviors, and annual energy consumption of residential buildings, and their distribution frequency in the U.S. and by region [63]. Data from 2005 and 2009 are used in this study to determine household member age distribution as this dataset has the information about the age of individual occupants in a household. Dataset of two different years data are used to evaluate the consistency in the analysis. 108 3.3. Survey Data To provide an initial assessment of this impact on household occupancy schedules, a survey was developed and distributed to a random sample of the Midwest residents (n = 3,165) between 7/10/2020 to 10/30/2020, administrated using a mixture of online and mail survey methods. The Midwest region was chosen because the median household income in this region is close to the national value [64]. A total of 415 responses (249 online and 166 mail responses) were collected, resulting in a response rate of 13.1%. After removing repetitions and empty entries, a final 376 valid responses resulted. After the data collection, the data was processed and the information with missing data was removed. The descriptive statistics of the final dataset are listed in Table 2. Table 4-2. Descriptive Statistics of Survey Sample Variable Mean Std. Dev. Min Max Age (n=309) 59 15,803 24 94 Household Income (est.) (n=293) 86,391 50,816 12,500 175,000 # of Household members 2 1.35 1 10 # Children (<18) (n=309) 0.5 0.94 0 6 # Adult (>18 & <65) (n=309) 1.3 1.15 0 6 # Senior (≥ 65) (n=309) 0.7 0.82 0 3 4. Methodology The ATUS and CPS linked data was divided into LIHs, MIHs (middle income households) and HIHs based on previously discussed income thresholds (Table 1). Next, activity data was extracted from the ATUS data for each survey participant. Activities were first mapped to the presence or absence of the person in their home. For most, but not all activities, a location was provided. For activities where locations were not provided, such as sleeping, grooming and personal activities, these were assumed to occur at home. An example of the mapping of activity and locations with respect to the various type of activities is included in Table 3. The activity-mapped location information was then represented using a binary value distribution, where 1 represents that the person was present at home, and 0 indicates the person was not present. From the ATUS data, initially the presence/absence-based binary location distribution was created across a 24-hour time scale to represent a day-long period. Next, the location information was subdivided into 5-minute time intervals, based on the minimum frequency of 5 minutes for 109 activities logged in the ATUS data collection process. As such, it was assumed that that during the 5-minute time period, the person was doing the same activity. Table 4-3. Sample of common activities mapped to the presence or absence of a person at home Example activities given in ATUS Presence in residential building Work, main job No (given) Eating and drinking Yes (given) Television and movies Yes (given) Washing, dressing oneself Yes (assumed) Socializing and communicating with others Yes (given) Sleeping Yes (assumed) From the ATUS dataset, age information was collected, and divided into seven groups, including age groups of under 25, 25 to 34, 35 to 44, 45 to 54, 55 to 64, 65 to 74, and over 75, to be consistent with the age distributions used in the RECS data [63]. The ATUS activity data was also divided into weekday and weekend subsets, to be consistent with previous literature that demonstrated significant differences in occupancy schedules on weekdays and weekends [65]. As a result of these divisions in age, weekday/weekend, and income groups, the ATUS data were split into 42 subgroups. Next, the average occupancy profiles for each subgroup were calculated and compared between income groups on both weekdays and weekends. Then cluster analysis was performed to identify variations in typical occupancy profiles for different income groups and household characteristics. The Dynamic Time Wrapping (DTW) [66] algorithm was used for cluster analysis, as each of the profiles can be represented as an individual time series. Dynamic time wrapping is pattern matching algorithm which was originally developed to perform speech recognition [67,68]. This method is used to identify patterns among two different sets of data. This is accomplished by considering two series X of size i and Y of length j where the time series can be represented by X = {x1, x2, x3, .... , xi} and Y = {y1, y2, y3, .... , yj}. For each element of m in X and of n in Y, the Euclidean distance between the points is evaluated using Equation 1. d( x+ , y0 ) = ( x+ − y0 )1 (1) A lower value of d represents a smaller distance between the two points. Next a path is identified which has a minimum cost function, where the cost function represents the summation of all the distance points. The advantage of the DTW method over other methods is that this method clusters 110 the profiles based on their pattern even when they were not time synchronized, i.e., there is some time lag between them [66]. Different studies have used DTW to evaluate the optimal alignment for time series data, such as for evaluating occupant behavior during driving or walking activities [69-74]. The variation in electricity consumption in a residential building based on occupant behavior is also studied using DTW method [75,76]. In this study ‘dtwclust’ package was used in R to implement this clustering algorithm [77]. DTW barycenter averaging (dba) is used as the centroid for the clustering, whereas global alignment kernels are used as the distance [78 - 80]. Based on our previous studies on ATUS data, 3 clusters were selected to analysis each of the subgroup data [62]. To evaluate the typical age distribution of household members, RECS 2005 and RECS 2009 were utilized. The RECS 2015 data was not used because it did not include the age of all household members. The percentage of households with different age group combinations (e.g., 2 people age 45-54 and 1 under 25) were then calculated. Using these values, the relative distribution for different household member age group combinations were evaluated, identifying the most common age distributions of household members for LIHs, MIHs and HIHs. Based on household demographics and the average occupancy profiles calculated, occupancy profiles were created for single and multi-person households based on the most common age distribution combinations. Finally, given the likely substantial impact COVID-19 pandemic on occupancy schedules and profiles, using survey data collected for a sample of 376 households in the Midwest region of the U.S. in 2020 on occupancy schedules during the pandemic. Compared with the general Midwest population, our sample of respondents were of relatively older age (average age =59), composed of more men than women, and had a higher household income (estimate = $86,391). Our sample of respondents also have a higher level of education, more retirees, more homeowners, and more people living in single-family homes. However, this data still provides an initial comparison of COVID-19 impacts and thus included in this work. The results of this survey were analyzed to provide a preliminary assessment of the impact of the pandemic on residential occupancy profiles. 111 5. Results and Discussion 5.1. Average Occupancy Profiles of Individuals Average occupancy profiles for the three income groups on both weekdays and weekends were calculated (Figure 1). For each 5-minute timestep, the variation in occupancy fraction is shown, representing the fraction of people in the home with respect to the total number of household members. Thus, an occupancy fraction of 1 represents all individuals being at home, whereas 0 implies none are present, across the dataset. Figure 4-1. Average occupancy profiles of individuals in HIHs, MIHs and LIHs on both weekdays and weekends As shown in Figure 1, the average occupancy fraction remains close to 1 in the middle of the night and early morning, then it begins to decrease. The average occupancy fraction reaches its minimum at around noon and then, depending on whether it is a weekday or weekend, it rises again from afternoon to evening. In general, irrespective of the income level of the individuals, the occupancy fraction was higher on weekends compared to weekdays, suggesting that people generally spent more time at home on weekends, compared to weekdays. In addition, during weekends, the occupancy fraction shifts slightly to the right compared to weekdays, which suggests that people generally leave their home later on weekends. Overall, on weekdays, the average occupancy fraction is 66%, 71% and 80% for high, middle and low-income groups respectively, whereas on 112 weekends they are 79%, 81% and 85%. Thus, irrespective of the day of the week, individuals in the low-income group spent the most time at home, on average, in comparison to the other groups, while the high-income group spent the least amount of time at home. This may be because many people in lower income households are individual females caring for children, which may require them to stay at home to care for their children [81]. One interesting finding from Figure 1 is that the average occupancy fraction of HIHs on weekends was higher than the weekday occupancy fraction of LIHs, which represents that for certain time of a day, HIH members spent less time at home on weekends, compared to those in LIHs on weekdays. The variation in occupancy fraction throughout the day for different age groups was analyzed and shown in Figure 2. The daily average occupancy fraction variation is also calculated for individuals in each of the three income groups and different age groups (Table 3). This shows that the occupancy fraction follows similar patterns on both weekdays and weekends across all age groups. However, on weekdays, the impact of age on occupancy fraction is much more significant compared to weekends. For HIHs, on weekdays, the daily minimum occupancy fraction varies from 19% to 58%, depending on the age group, whereas for MIHs and LIHs, these values are 34% to 56% and 41% to 75%, respectively. However, on weekends, the minimum occupancy fraction values vary from 50% to 61%, 55% to 64% and 58% to 72% respectively for HIHs, MIHs and LIHs. It is also noted that on weekends, the minimum occupancy fraction occurs for a much smaller time span compared to weekdays. 113 Figure 4-2. Average occupancy fraction variation for individuals in (a) LIHs, (b) MIHs and (c) HIHs by age group on both (1) weekdays and (2) weekends Table 4-4. Average daily occupancy fraction for LIH, MIH, and HIH individuals for different age groups on weekday and weekends Day of week Income status Under 25 25-34 35-44 45-54 55-64 65-74 Over 75 HIH 65.5 59.8 61.5 61.9 66.2 75.2 83.0 Weekday MIH 68.5 68.0 67.4 67.9 71.2 77.7 80.6 LIH 72.6 73.7 76.1 79.6 85.1 88.9 91.7 HIH 77.8 76.9 78.2 78.2 78.8 81.4 85.1 Weekend MIH 79.9 79.9 79.6 79.9 82.0 84.1 85.9 LIH 81.7 81.7 83.1 85.9 88.3 89.6 91.4 From Table 4, it can be seen that the daily average occupancy fraction varies among age groups, with the variation following similar trends on both weekdays and weekends. For HIHs, the daily average occupancy fraction was lowest for those 25 to 34, then increased with an increase in age. The daily average occupancy fraction decreases with an increase in age group until the 35 to 44 age group, and then it increases with increasing age. For LIHs, the daily average occupancy 114 fraction rises with an increase in age for both weekdays and weekends. This suggests the importance of age in evaluating occupancy scenarios in the United States. 5.2. Cluster Analysis of Individual Profiles of Occupancy Schedules Following the analysis of the average occupancy fraction, cluster analysis was performed using DTW on each of the 42 subset of data which was created based on the variables discussed in the methodology section. As mentioned, a total of three clusters were selected for each subset. As a result of cluster analysis of all subdivisions, in total 42 X 3 = 126 profiles were obtained. The resulting profiles were studied individually, where the characteristics of the profiles i.e., when an individual leaves the space and the time duration of the absence, were evaluated. These characteristics were compared across all the obtained profiles. The average of each of the clusters was also evaluated. Significant similarities were seen among multiple profiles. Thus, based on these characteristics and the average profiles, the resulted 126 cluster profiles were categorized in 9 major types of schedules. These profiles are shown in Figure 3. These nine profiles are named as follows: (a) Stay home, (b) Day absence, (c) Day absence, less time, (d) Extended absence, (e) Day absence, earlier in day, (f) Day absence, later in day, (g) Absence very early in day, (h) Absence very late at night and (i) Night absence. We note that our analysis has not distinguished whether people were away from home for work or for other purposes and activities. The specific activities conducted outside the home were not the focus in this research, however, this would be a beneficial topic for future work. Of these profiles, Stay home was the profile where occupants spent more time at home. Day absence represents those that left home in the morning at approximately 7 to 8 am and returned home in the evening around 6 pm. For those who were away from home but for a shorter period (3 to 4 hours), this profile is termed Day absence less time. Conversely, those away from home for extended periods, leaving their home early in the morning and returning home late at night, followed an Extended absence profile. Day absence, earlier in day and Day absence, later in day represent profiles when people left home in the early morning and afternoon, respectively, and were not present for a 4 to 8 hour time span. The Absence very early in day profile represents where people were not present in their home during the early morning, specifically from 2 to 3 am to around 10 to 11 am, whereas Absence very late at night was when individuals were at home in 115 the evening and late at night. The last profile was Night absence, which represents when individuals were not present at home throughout the night and stayed home during the day. To analyze the influence of individuals’ demographics on occupancy patterns, we also evaluated the percentage of people belonging to each of income level and age groups who followed these different types of schedules, the results of which are shown in Table 5(a) and 4(b), respectively, for weekdays and weekends. These tables show the three most common types of occupancy profiles (from Figure 3) for each subset of individuals, by age, income and weekday/weekend. Figure 4-3. Occupancy profiles resulting from cluster analysis, including (a) Stay home, (b) Day absence, (c) Day absence less time, (d) Extended absence, (e) Day absence earlier in day, (f) Day absence later in day, (g) Absence very early in day, (h) Absence very late at night end, and (i) Night absence 116 Table 4-5. Most common occupancy schedule types and percentage of individuals these schedules by income group and age on (a) weekdays and (b) weekends (Note: bold text indicates the cluster with the greatest percent of people) As shown in Table 5(a), on weekdays for LIHs, irrespective of age group, the majority of people follow the Stay home schedule. For MIHs, those over the age of 55 mostly also follow the Stay home schedule, whereas people under 55 generally follow schedules other than Stay home. For 117 HIHs, the majority of people under 65 follow one of the schedules where they are away from home for some period of time, with the Day absence schedule being most common. For HIHs, the Extended absence profile also appears to be more common among the various age groups, however, in general, profiles obtained for HIHs are more consistent across all age groups compared to others. For LIHs, however, a wide range of profile types can be seen on weekdays that represents a higher uncertainty in schedule for lower income individual, as was mentioned by Golden [82]. On weekends, across all income and age groups, more than half of people followed Stay home profiles or they leave home for only very short periods of time. The Day absence less time profile was also very common for occupants in HIHs whereas for LIHs a variety of profile types can be seen. Similar to weekdays, this represents that people in LIHs have much more variations in their daily schedules compared to HIHs [83, 84]. However, at the same time, household income has more impact on their occupancy profiles on weekdays compared to weekends. 5.3. Household-Level Occupancy Schedules in the U. S. In addition to analyze individuals’ occupancy profiles, this research also focuses on household- level occupancy profiles for considering the potential energy savings from the use of smart technologies that would adjust energy consuming devices during non-occupied times. The most common household sizes in this sample are 1-, 2-, 3-, and 4-member households according to RECS data [85]; these size households represent more than 90% of all households in the U.S. For the purpose of analysis, each age group was coded numerically in increasing order of age, as shown in Table 6. Next, the number of households within each of the age group combinations was calculated and their relative distributions were evaluated. Table 4-6. Age groups coding for household-level occupancy schedule analysis Age code Age group 1 Under 25 2 25 to 34 3 35 to 44 4 45 to 54 5 55 to 64 6 65 to 74 7 Over 75 118 The distribution of age ranges of 1-member households is shown in Figure 4. As shown in Figure 4, both the RECS 2005 and 2009 datasets show similar distributions. For LIHs, over 75 age group is the most common, representing 24-29% of all 1-member households. For HIHs, the 45 to 54 and 55 to 64 age groups are the most common, representing approximately 50% of single-member households. For MIHs, the most common age distribution varied across RECS 2005 and 2009 dataset where over 75 and 55 to 64 are the two most common. The over 75 age group is a significantly lower percentage representation for the HIHs. This is likely due to household members being retired at this age, and thus having lower incomes as a result. Figure 4-4. Percent of 1-member households by age range for (a) LIHs (b) MIHs and (c) HIHs; Note: age categories are arranged in increasing percentage representation (smallest at the top, largest at the bottom) Similar results have also been studied for 2-, 3-, and 4-member households. The percentage of households by household composition (i.e. age range of household members) is shown in Figures 5 through 7. For example, Figure 5(a) shows the distribution of age range combinations for LIHs, from least common (top) to most common (bottom). ‘6,6’ indicates that there are two people with the age code ‘6’, or 65 to 74 years old (see Table 6); similarly, ‘1,1’ indicates there are two people with the age code of ‘1’ or under 25 years old. This naming scheme is used consistently throughout this research and the 3- and 4-member figures as well. The age composition of 2-member households by income levels is shown in Figure 5. Only the top 11 most common compositions are shown to highlight the most common age combinations, representing 70%-90% of households of this size. For LIH, the primary age combinations were younger people, where both are either under 25 years, or one under 25 years and the other 25 to 34. For MIHs, approximately 25% of households are in older age groups, ranging from 55 to 74 119 years old. The age group distributions of HIHs are similar to MIHs, where around 35% of the households include people 45 to 64 years old. Figure 4-5. Percent of 2-member households by age range for (a) LIHs (b) MIHs and (c) HIHs (Note: Codes used on the y-axis indicate the age code (Table 6) for each household member) The age composition of 3-member households is shown in Figure 6, for the top 11 most common combinations, representing 70-80% of households. For LIHs, the top 4 most common household age combinations (45%-60% of households) including 2 household members in the youngest age group, paired with one older household member. This may represent single parents of two children, two parents of different ages with a child, or roommate scenarios. Similarly, four additional household types (21-25% of households) include 1 household member under 25 and two older adults, likely the child of the two older adults. For MIHs, approximately 32% of households, i.e. the top three most common combinations, have two people ages 25 to 44 years, with the third person is under 25 years, likely the child of the two adults. For HIHs, the most common age groups are slightly higher compared to MIHs, where 2 are 35 to 54 years and the other is under 25 years. This may be because those who delay having children until an older age are more likely to have high household incomes than those that have children earlier in their career [86, 87]. Similar results are shown in Figure 7 for 4-member households. Households with younger members are more common, where for around 40%, all household members are 34 or younger. 120 For HIHs, more than 50% of the households have 2 people ages 35 to 54, and the two others are under 25; this likely represents two working adults and their two children. Similarly, for MIHs approximately 45% have 2 adults ages 25 to 54, and members under 25. For LIHs, 39-43% of households are made up of members 34 years old and younger. Figure 4-6. Percent of 3-member households by age range for (a) LIHs (b) MIHs and (c) HIHs (Note: Codes used on the y-axis indicate the age code (Table 6) for each household member) Figure 4-7. Percent of 4-member households by age range for (a) LIHs (b) MIHs and (c) HIHs (Note: Codes used on the y-axis indicate the age code (Table 6) for each household member) 121 Combining these results, the most common age combinations for each household size are summarized in Table 7 by income level. HIHs more commonly included members ages 45 to 54, irrespective of the number of household members. For MIHs, 1- and 2- member households generally have older members ages 55 to 64, but for 3- and 4-member households, the household members are generally in younger age groups. For LIHs, 1-member households were generally those over 75; 2-, 3- and 4-member household members were significantly younger, where in most combinations, there is at least one or more members under 25. Table 4-7. Most common household age combinations by household size and age ranges Number of household members Household income level Household member ages LIH Over 75 1-member household MIH 55 to 64 HIH 45 to 64 LIH Both under 25 2-member household MIH Both 55 to 64 HIH Both 45 to 64 LIH Two under 25; one 25 to 44 3-member household MIH Two 45 to 54; one under 25 HIH Two 45 to 54; one under 25 LIH Three under 25; one 25 to 34 4-member household MIH Two under 25; two 35 to 44 HIH Two under 25; two 35 to 54 Using the most common age range combinations (average percentage across both 2005 and 2009 RECS data) for each size household at each income level, the average occupancy fraction for weekdays and weekends is shown in Figure 8. For all households, regardless of the number of household members, the occupancy fraction of HIHs (black) is significantly lower during the daytime compared to that of LIHs (orange), which are lowest among the three income levels. The maximum occupancy fraction, generally occurring at night, for households across all three income groups on both weekdays and weekends are similar. However, on weekdays, the minimum occupancy fraction for 2-, 3- and 4-member LIHs are approximately 0.2 higher compared to that of HIHs. The difference in the minimum occupancy fraction is even higher for 1-member LIHs, where there is an 0.47 difference compared to the HIHs. On weekends, the differences in the occupancy fraction across LIHs and HIHs are lower compared to weekdays. For 2, 3 and 4-member 122 households, this difference reduces to 0.07, whereas for 1-member households the difference is 0.18. The average occupancy fraction of MIHs (blue) is in between the HIHs and LIHs. Figure 4-8. Occupancy fraction for the most common age combinations for (a) 1-member, (b) 2- member, (c) 3-member and (d) 4-member households for LIHs, MIHs, and HIHs on weekdays and weekends As it is shown in the figure, irrespective of income level, the occupancy fraction begins close to one in the morning and then begins to decrease. The occupancy fraction reaches its minimum at around noon and then increases slowly, reaching close to one in the evening. For HIHs, the occupancy fraction begins to decrease earlier compared to LIHs. In addition, the minimum occupancy fraction occurs for a longer time span in HIHs compared to LIHs. The difference in starting time among HIHs and LIHs is smaller on weekends compared to weekdays. 5.4. Impact on Energy Performance The impact on the energy performance due to the variation in occupancy schedules for different income groups was then evaluated. A prototype residential building model developed by the 123 Pacific Northwest National Laboratory was used in this study as the baseline building model [88]. To evaluate the impact of climate conditions, two extreme climates were selected, including Houston (Climate Zone 2A, hot/humid) and Minneapolis (Climate Zone 6A, cold/humid) [89]. To evaluate the impact of energy savings due to occupant-based controls, demo schedules were created. For each income group, both the weekday and weekend average schedules were studied. Occupancy fraction of 0.5 was selected as the threshold, where higher than 0.5 represent presence of occupant. Similarly, if the occupancy fraction is below 0.5, it was assumed that there is no occupant in household. A temperature setback was used for occupancy-based controls to estimate the impact on energy performance of the varied occupancy. Heating and cooling setpoint temperatures of 21.7°C and 23.7°C, respectively were used when a building is occupied (occupancy fraction > 0.5); when unoccupied (occupancy fraction < 0.5), a setback of 2°C and 6°C for both heating and cooling are used. These were stored as .CSV files, and imported into the EnergyPlus model, then simulated using EnergyPlus v.9.5 [90]. The resulting heating, cooling, and total HVAC energy consumption and percentage change with respect to the baseline model are shown in Figure 9. (a) Houston Houston (b) Minnesota 30 30 30 25 25 25 Percentage savings Percentage savings Percentage savings 20 20 20 15 15 15 10 10 10 5 5 5 0 0 0 HIH2 HIH6 MIH2 MIH6HIH2 LIH2 HIH6 LIH6MIH2 MIH6 HIH2LIH2 HIH6LIH6 MIH2 MIH6 LIH2 LIH6 Heating Cooling Total Heating Cooling Total Heating Cooling Total Figure 4-9. Percentage annual HVAC energy savings due to the use of setbacks for occupant- based control for (a) Houston and (b) Minneapolis across income groups (Note: HI = high income, MI = middle income, LI = low income; 2 = 2°C setback when unoccupied; 6 = 6°C setback when unoccupied) The energy savings is substantially higher (1-17%) when using the larger setbacks (6°C) as compared to the smaller setbacks (1-12%), which is expected. Household income also had a substantial impact on savings. In Houston, 18% and 14% HVAC energy savings can be saved for high- and middle-income households respectively, whereas for low-income households, energy savings is estimated as 1%. A similar pattern was found in Minneapolis, with an HVAC energy savings of approximately 10% for high- and middle-income households, but only 1% for low- 124 income households. This initial analysis shows the importance of considering the household income when evaluating energy savings potential of various energy efficiency solutions, as well as the importance of having occupancy profiles that are specific to household income groups. More detailed study of occupancy scenario is needed to accurately evaluate the energy performance of building systems. 5.5. Impact of COVID-19 on Occupancy Scenarios We note that the data discussed in this analysis in prior sections is based on historic data collected prior to the COVID-19 pandemic, which significantly disrupted the typical occupancy scenarios of many households in the U.S. As mentioned by Brynjolfsson et al. (2020) around 35.2% of workers switched to work from home due to COVID-19 [91], which is significantly higher than the non-COVID-19 scenario [92]. As schools were generally closed, children were also at home more and generally attended classes remotely from home [93, 94]. In addition, with travel restrictions and event and gathering cancellation, among other outside-the-home events cancellations, more people stayed at home more, based on analysis of cell phone data [95]. We note that this sample of respondents are older (and thus likely more retirees), their household size is smaller, and they have higher household incomes. Similar to the ATUS data division, survey results were divided based on income level and number of household members (Table 8). Around 50% of the participants are in HIHs, and 40% are from MIHs. Participants belonging to LIHs consisted of only 10% of the overall survey responses. For this reason, the data from LIHs for 2, 3 and 4-member households were excluded from the analysis as they represent less than 2% of the data. Among the HIHs, more than 50% of the participants belonged to the 2-member household, whereas for MIHs, that number is around 40%. Thus, while the survey does not represent the household demographics of the United States, it does show a preliminary analysis the relative impact of COVID-19 on occupants’ presence in residential buildings. 125 Table 4-8. Distribution of household characteristics of the survey Household income level Number of household members Percentage of participants 1-member 5.5 2-members 2.0 LIH 3-members 0.8 4-members 0.8 1-member 9.0 2-members 16.5 MIH 3-members 8.2 4-members 6.3 1-member 10.6 2-members 27.8 HIH 3-members 5.9 4-members 6.7 The survey asked participants whether their residences are “always occupied”, “sometimes occupied”, or “not occupied” at different hours of the day from 6 am to 10 pm for a typical day on a weekday and weekend. Along with their current schedule during the COVID-19 pandemic scenario (CS), participants were also asked to predict the occupancy scenario of their household for the same timespan when the COVID-19 pandemic is over (PCS). This occupancy information is mapped to the income levels of the households and the number of household members. The variation in the resulting occupancy profiles is shown in Figure 10. This shows the occupancy fraction during COVID-19 is higher throughout the day for both weekdays and weekends for all household types. As expected, the predicted occupancy fraction for the post COVID-19 scenarios decreases slightly compared to the COVID-19 scenarios, which means people would spend less time in home. As shown in Figure 10.a1, for 1-member households on weekdays, the occupancy fraction of both HIHs and LIHs are similar and close to 1. The occupancy fraction for MIHs is comparatively lower which suggests that people in MIHs are in their home less during the pandemic compared to the other groups. This may be because HIH adults may work in jobs that do not require in-person work (able to work remotely), and thus they work from home for most or all of the day, compared to MIHs which may be in professions that require more in-person work [96]. The lower income group is the most affected income group, where a significant portion are likely facing unemployment challenges [97]. Such challenges may result in the comparatively higher occupancy fractions in LIHs. According to the post-pandemic predicted schedule results from the survey, the occupancy fraction of MIHs is the lowest, followed by HIHs. It is the highest 126 for LIHs. Similar trend can also be seen for 2-, 3- and 4-member households, where the occupancy fraction of HIHs are higher than MIHs during COVID-19. Overall, trends in this data suggest the time spent in the residential buildings will reduce post pandemic, as expected. Figure 4-10. Survey-based impact of COVID-19 on the household occupancy profiles for (a) 1- member, (b) 2-member, (c) 3-member and (d) 4-member households for (1) weekdays and (2) weekends (Note: 2,3 and 4-member LIHs are not shown due to limited data; CS = During COVID; PSC = Post-COVID) 127 The variation in occupancy schedules across different income groups irrespective of the number of household members for both weekdays and weekends is shown in Figure 11. On weekends there is a significant difference in the occupancy fraction across income groups for both current and predicted future scenario. However, on weekdays, the occupancy varies significantly for the post- pandemic scenario. A significant drop in occupancy fraction can be seen for MIHs during the daytime. The reduction in occupancy fraction for HIHs is lower compared to MIHs. This may be because during the post pandemic period, a significant amount of the workforce may consider work-from-home scenarios, thus impacting the occupancy fraction [98, 99]. The variation in predicted occupant fraction of households is comparatively less for LIHs. More information for LIHs is needed to analyze the impacts of the pandemic on the occupants belong to this income group. Figure 4-11. Impact of COVID-19 on the HIH, MIH and LIH household occupancy profiles for (a) weekdays and (b) weekends (Note: CS = During COVID; PSC = Post-COVID) 128 6. Conclusions In this study, ATUS and CPS data were analyzed to study the occupancy profiles for individuals and households of 1- to 4-members in the United States. These individuals and households are divided into three income groups, including low-income, medium-income and high-income households, and 7 age groups. Occupancy profiles were also studied for both weekdays and weekends separately. The average occupancy profiles were evaluated and then cluster analysis was implemented using these profiles. After this, the typical age combinations of households by household size were determined based on RECS data, resulting a comparison of occupancy profiles for the most common age combination by income level. Finally, the impact of the COVID- 19 pandemic on occupancy profiles for different size households was estimated based on collected survey data. Overall, the key findings of this study are as follows: • The average occupancy fraction of individuals in older age groups is higher compared to individuals in younger age groups. In addition, the average occupancy fraction of individuals on weekends is higher compared to weekdays. On weekdays, the minimum occupancy fraction for older age groups is approximately 0.3 higher compared to younger age groups. On weekends, the difference is smaller, approximately 0.1 among different age groups. • The average occupancy fraction of LIH individuals is significantly higher compared to MIH and HIH individuals. In some scenarios, specifically 10 am to 2 pm, the average occupancy fraction of HIH individuals on weekends is lower compared to that of weekdays in LIH individuals. Overall, midday, the average occupancy fraction of LIH is approximately 0.3 and 0.1 higher for weekdays and weekends, respectively, compared to HIHs. The occupancy fraction of MIH varies between these two profiles. • In total, 9 different types of occupancy profiles were obtained as a result of the cluster analysis, including (a) Stay home, (b) Day absence, (c) Day absence, less time, (d) Extended absence, (e) Day absence, earlier in day, (f) Day absence, later in day, (g) Absence very early in day, (h) Absence very late at night and (i) Night absence. Among these profiles, the Stay home profile was the most common for individuals on weekends. On weekdays, the Stay home profile is only the most common for LIH individuals, unlike HIHs, where Day absence or Extended absence profiles were most common. For MIH 129 individuals, younger age groups generally followed one of the absence schedules, and the older age groups followed the Stay home schedule. • The occupancy fraction of households varies with respect to the number of household members and their overall household income. On weekdays, the occupancy fraction of the most common HIHs for 2-, 3- and 4-member households are 0.2 lower than the LIH. This difference is as high as 0.47 for 1-member households. The difference reduces on the weekend, from 0.07 to 0.18. • Due to the impact of COVID-19, the occupancy fraction of households is reported to have increased significantly throughout a typical weekday and weekend, with some variation by income level. There are some limitations to this study. Some of the datasets used are self-reported data which is subject to human error. In addition, the ATUS survey represents occupancy data for one particular day, which may be different from the specific occupant’s average profile. Both the surveys also do not provide information about the correlation of schedules among different occupants in a household. More detailed information about the long-term activity of households and collective occupancy scenario for entire households is needed for more accurate analysis. The result of this study can be used to estimate average occupancy schedules for individuals based on their income level and age. These profiles, mapped to the household level, also provide a predicted occupancy schedule at the household level, by household size and income. These can be used as input into residential building energy simulation scenarios to assess income and household size impacts on building energy performance, in particular for devices that control buildings systems such as lighting and HVAC systems based on whether or not the building is occupied. Overall, the data suggests that middle- and high-income households, and households with working age or younger individuals can benefit more from energy savings resulting from the use of such systems since these households overall have more extended unoccupied periods. By contrast, based on the results of this study, lower income households and those with members in older age groups can benefit less from energy savings associated with such controls as these households are generally at home more often. 130 With additional linked data on how such households use energy consuming devices, additional impacts of occupancy on energy consumption can also be studied. The results of this study can also be used in future work to create a stochastic occupancy simulator for different types of residential buildings in the United States. 7. Acknowledgements The information, data, or work presented herein was funded in part by the Advanced Research Projects Agency-Energy (ARPA-E), U.S. Department of Energy, under Award Number DE- AR0001288. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof. The author also acknowledges the support of the Sloan Foundation. 131 REFERENCES [1] United States Energy Information Administration (US EIA) https://www.eia.gov/totalenergy/data/monthly/ (Accessed October 20th, 2020) [2] United States Energy Information Administration (US EIA). 2020. International Energy Outlook, https://www.eia.gov/outlooks/aeo/ (Accessed October 20th, 2020) [3] United Nations Environment Program (UNEP). 2016. Sustainable building and climate initiative https://energies2050.org/sustainable-buildings-and-climate-initiative-unep- sbci/?lang=en (Accessed October 20th, 2020) [4] United States Environmental Protection Agency. 2014. Sources of greenhouse gas emissions. EPA headquarters, Washington, D.C., U.S.A. https://www.epa.gov/ghgemissions/sources- greenhouse-gas-emissions (Accessed October 20th, 2020) [5] Core Writing Team, Pachauri, R. K., and Meyer, L. A. (eds.). 2002. “IPCC, 2014: Climate Change 2014: Synthesis Report. Contribution of Working Groups I, II and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change.” IPCC 40 (1): 151. https://doi.org/10.1046/j.1365-2559.2002.1340a.x. [6] U.S. Energy Information Administration. 2015, “A look at the U.S. Commercial Building Stock: results from EIA's 2012 Commercial Buildings Energy Consumption Survey (CBECS)”, Release date: March 4, 2015. https://www.eia.gov/consumption/commercial/reports/2012/buildstock/. [7] Hong T., D’Ocaa S., Turnera W.J.N., Taylor-Lange S.C. 2015. “An ontology to represent energy-related occupant behavior in buildings. Part I: introduction to the DNAs framework” Build. Environ., 92, 764-777. [8] Yoshino H., Hong T., Nord N. 2017. “IEA EBC annex 53: total energy use in buildings— analysis and evaluation methods”, Energy Build 152, 124–136, doi: 10. 1016/j.enbuild.2017.07.038 [9] Yan D., Brien W.O., Hong T., Feng X., Gunay H.B., Tahmasebi F. 2015. “Occupant behavior modeling for building performance simulation: Current state and future challenges”, 107, 264– 278. doi:10.1016/j.enbuild.2015.08.032. [10] Yu Z., Fung B.C.M., Haghighat F., Yoshino H., Morofsky E., 2011. “A systematic procedure to study the influence of occupant behavior on building energy consumption”. Energy Build, 43, pp. 1409- 1417, 10.1016/j.enbuild.2011.02.002. [11] Lo L.J., Novoselac A., 2010. “Localized air-conditioning with occupancy control in an open office”. Energy Build., 42 (7), pp. 1120-1128. [12] Li, Z.-J., Jiang, Y., Wei, Q.-P., 2014, “Survey on energy consumption of air conditioning in summer in a residential building in Beijing”, Heat. Vent. Air Cond., 37 (2007), pp. 46-51. 132 [13] An J, Yan D, Hong T. 2018. “Clustering and statistical analyses of air-conditioning intensity 457 and use patterns in residential buildings[J]”. Energy and Buildings, 174: 214-227. [14] Xia, D., Lou, S., Huang, Y., Zhao, Y., Li, D. H., & Zhou, X. 2019. “A study on occupant behaviour related to air-conditioning usage in residential buildings”. Energy and Buildings, 203, 109446. [15] Santin, O. Guerra, Itard, L., Visscher, H., 2009. “The effect of occupancy and building characteristics on energy use for space and water heating in Dutch residential stock”, Energy Build, 41, pp. 1223-1232, 10.1016/j.enbuild.2009.07.002. [16] Wilke, U. 2013, “Probabilistic Bottom-up Modelling of Occupancy and Activities to Predict Electricity Demand in Residential Buildings”. PhD Thesis. 5673 172, 10.5075/epfl-thesis-5673. [17] Popoola, O. M. 2018. “Computational Intelligence Modelling Based on Variables Interlinked with Behavioral Tendencies for Energy Usage Profile – A Necessity.” Renewable and Sustainable Energy Reviews 82: 60–72. https://doi.org/10.1016/j.rser.2017.09.020. [18] Dong, B., Li, Z., & Mcfadden, G. 2015. “An investigation on energy-related occupancy behavior for low-income residential buildings. Science and Technology for the Built Environment”, 21(6), 892-901. [19] Kolokotsa D. and Santamouris M., 2015. “Review of the indoor environmental quality and energy consumption studies for low-income households in Europe.,” Sci. Total Environ., vol. 536, pp. 316–330. [20] Xu X and Chen C., 2018. “Energy efficiency and energy justice for U. S. low-income households : An analysis of multifaceted challenges and potential,” Energy Policy, vol. 128, June 2018, pp.763–774. [21] Drehobl A., Ross L., and Ayala R. 2020. “Report: Low-Income Households, Communities of Color Face High ‘Energy Burden’ Entering Recession,” ACEEE, 2020. [Online]. Available: https://www.aceee.org/press-release/2020/09/report-low-income-households-communities-color- face-high-energy-burden [22] Santos, C. M. D. 2020. "Energy-efficient technology retrofit investment behaviors of households in lower and higher income regions." [23] Deru, M., Field K., Studer D., Benne K., Griffith B., Torcellini P., Liu B., et al. 2011. “U.S. Department of Energy commercial reference building models of the national building stock.” Nrel/Tp-5500-46861, no. February 2011: 1–118. [24] Goel, Rosenberg, Xie Athalye, Hart Wang, and Mendon Zhang. Enhancements to ASHRAE standard 90.1 prototype building models. No. PNNL-23269. Pacific Northwest National Lab. (PNNL), Richland, WA (United States), 2014 [25] ASHRAE 90.1 User's Manual ANSI/ASHRAE/IESNA Standard 90.1-1989, American Society of Heating, Refrigerating and Air-Conditioning Engineers, Atlanta, GA (1989) 133 [26] Advanced Energy Design Guides, 2018. (accessed October 10, 2020) [27] BEopt, 2019, http://beopt.nrel.gov/. (accessed October 10, 2020) [28] Christensen, C., Anderson, R., Horowitz, S., Courtney, A., and Spencer, J. 2006v“BEopt (TM) software for building energy optimization: features and capabilities”. United States: N. p. Web. 10.2172/891598. [29] S. D’Oca, T. Hong, 2015. “Occupancy schedules learning process through a data mining framework”, Energy Build., 88, pp. 395-408 [30] M. Eguaras-Martínez, M. Vidaurre-Arbizu, C. Martín-Gómez, 2014. “Simulation and evaluation of building information modeling in a real pilot site”, Appl. Energy. 114, 475–484, doi: 10.1016/j.apenergy.2013.09.047. [31] Saha, H., Florita, A. R., Henze, G. P., Sarkar., S., 2019, “Occupancy sensing in buildings: A review of data analytics approaches”, Energy&Buildings188–189, 278–285 [32] Page, J., Robinson, D., Morel, N. and Scartezzini. J. L. 2008. “A Generalised Stochastic Model for the Simulation of Occupant Presence.” Energy and Buildings 40 (2): 83–98. [33] Richardson, I, Thomson M., and Infield D. 2008. “A High-Resolution Domestic Building Occupancy Model for Energy Demand Simulations.” Energy and Buildings 40 (8): 1560–66 https://doi.org/10.1016/j.enbuild.2008.02.006 [34] Widén, J., and Wäckelgård E. 2010. “A High-Resolution Stochastic Model of Domestic Activity Patterns and Electricity Demand.” Applied Energy 87 (6): 1880–92. [35] López-Rodríguez, M.A., Santiago, I., Trillo-Montero, D., Torriti, J., Moreno-Munoz, A. 2013. “Analysis and modeling of active occupancy of the residential sector in Spain: an indicator of residential electricity consumption”, Energy Policy, 62, pp. 742-751, 10.1016/j.enpol.2013.07.095 [36] Vale, Raúl Fraguela, Juan José Lorenzo Castiñeiras, and Lara Varela Garrote. 2011. "Escuela, familias 552 y ocio en la conciliación de los tiempos cotidianos de la infancia." Revista de Investigación Educativa 29, no. 2: 429-446. [37] Adamopoulou, A. A., Tryferidis, A. M., Tzovaras, D. K. 2016. “A context-aware method for building occupancy 477 prediction.” Energy and Buildings 110, 229–244. [38] Wang, W., Chen, J, Song, X, 2017. “Modeling and predicting occupancy profile in office space with a Wi-Fi probe-based Dynamic Markov Time-Window Inference approach”, Building and Environment 124, 130-142 [39] Chen D., Barker, S., Subbaswamy, A., Irwin, D., Shenoy, P, 2013. "Non-intrusive occupancy monitoring using smart meters." In Proceedings of the 5th ACM Workshop on Embedded Systems For Energy-Efficient Buildings, pp.1-8. 134 [40] Lam, K. P., Höynck, M., Dong, B., Andrews, B., Chiou, Y., Zhang, R., Benitez, D., and Choi, J., 2009. "Occupancy detection through an extensive environmental sensor network in an open- plan office building." IBPSA Building Simulation 145: 1452-1459. 493 [41] Amato, G., Carrara, F., Falchi, F., Gennaro, C., Meghini, C., Vairo, C., 2017. "Deep learning for decentralized parking lot occupancy detection." Expert Systems with Applications 72: 327- 334. [42] Chen, Z., Zhao, R., Zhu, Q., Masood, M. K., Soh, Y. C., and Mao, K., 2017. "Building occupancy estimation with environmental sensors via CDBLSTM." IEEE Transactions on Industrial Electronics 64, no. 12: 9549-9559. [43] Zhao, J., Bertrand L., Lam, K. P., Yun, R., and Loftness, V., 2014. "Occupant behavior and schedule modeling for building energy simulation through office appliance power consumption data mining." Energy and Buildings 82: 341-355. 490 [44] Buttitta, G., William JN T., and Donal F., 2016. "Clustering of household occupancy profiles for archetype building models." Energy Procedia. [45] Vazquez, F. I., and Kastner W. 2011. “Clustering Methods for Occupancy Prediction in Smart Home Control.” Proceedings - ISIE 2011: 2011 IEEE International Symposium on Industrial Electronics, 1321–28. [46] Hailemariam, E., Goldstein, R., Attar, R., and Khan., A., 2011. “Real-time occupancy detection using decision trees with multiple sensor types.” In Proceedings of the 2011 Symposium on Simulation for Architecture and Urban Design (SimAUD '11). Society for Computer Simulation International, San Diego, CA, USA, 141-148 [47] Kleiminger, W., Beckel, C., Staake, T., and Santini. S., 2013. “Occupancy Detection from Electricity Consumption Data. In Proceedings of the 5th ACM Workshop on Embedded Systems For Energy-Efficient Buildings” (BuildSys'13). ACM, New York, NY, USA, Article 10 , 8 pages. DOI=http://dx.doi.org/10.1145/2528282.2528295 [48] Amayri, M, Arora A., Ploix S., Bandhyopadyay S., Ngo Q.D, and Badarla V.R. 2016. "Estimating occupancy in heterogeneous sensor environment." Energy and Buildings 129: 46-58. [49] Schwartz, Z., Uysal, M., Webb, T., Altin, M., 2016. “Hotel daily occupancy forecasting with competitive sets: a recursive algorithm”. Int. J. Contemp. Hosp. Manag., 28 (2), pp. 267-285. [50] Parker, K., J. Menasce Horowitz, and R. Minkin. "How the Coronavirus Outbreak Has–and Hasn’t–Changed the Way Americans Work”, Pew Research Center, December 9th, 2020. [51] “The never normal. Defining new ways of working beyond COVID-19”, Accenture, October 26, 2020. [52] Lee, J., “Five ways to imagine a better future of work in the post-covid era”, Forbes Human Resources Council, December 14, 2020 135 [53] Louie, D., “How COVID-19 pandemic forever changed the workplace, what companies are doing going forward”, ABC7NEWS, March 17, 2021 [54]. Weekes, W., “13 ways the coronavirus pandemic could forever change the way we work”, CNBC, April 29, 2020. [55]. Mudditt, J., “How offices will change after coronavirus”, BBC, May 14, 2020 [56]. Davenport, T. H., Redman, T. C., “Experiments and data for post-COVID-19 work arrangements”, MITSloan Management Review, March 23, 2021 [57]. “How the COVID-19 pandemic is changing the way we work”, Corehealth Technologies, Jul 16, 2020 [58] U.S. Bureau of Labor Statistics, American Time Use Survey, 2017. http:// www.bls.gov/tus/#data (Accessed on 10th May, 2020.). [59] U.S. Bureau of Labor Statistics, Current Population Survey, https://www.bls.gov/cps/ (Accessed on 10th May, 2020.) [60] Office of ASPE, U.S. Department of Health & Human Services, 2018. Poverty Guidelines. https://aspe.hhs.gov/poverty-guidelines. [61] Kochhar, R., Fry R., and Rohal M. 2016. "America’s Shrinking Middle Class: A Close Look at Changes Within Metropolitan Areas." Pew Research Center. [62] Mitra, D, Chu, Y, Cetin, K, Chen C, 2021. “Characteristics of residential occupancy profiles for different income groups in the United States”, ASHRAE winter conference, Chicago, Illinois, January 23- January 27, 2021 [63] U.S. Energy Information Administration, Residential Energy Consumption Survey, (2015). [64] American Community Survey 2018, United States Census Bureau, https://www.census.gov/acs/www/data/data-tables-and-tools/data-profiles/2018/, Accessed on March 1, 2021 [65] Balvedi, B. F., Ghisi, E, and Lamberts, R. 2018. “A Review of Occupant Behaviour in Residential Buildings.” Energy and Buildings 174: 495–505. [66] Berndt, D.J., and Clifford. J, 1994. "Using dynamic time warping to find patterns in time series." In KDD workshop, vol. 10, no. 16, pp. 359-370. [67] Sakoe, H. and Chiba, S. 1978. "Dynamic Programming Algorithm Optimization for Spoken Word Recognition," in IEEE Transactions on Acoustics, Speech and Signal Processing In Acoustics, vol 26, no. 1, pp. 43–49. 136 [68] Lawrence, R., Rosenberg, A., and Levinson. S. 1978. "Considerations in dynamic time warping algorithms for discrete word recognition." IEEE Transactions on Acoustics, Speech, and Signal Processing 26, no. 6: 575-582. [69] Saiprasert, C., Pholprasit, T., and Pattara-Atikom, W. 2013. "Detecting driving events using smartphone." In Proceedings of the 20th ITS World Congress, vol. 11, p. 11. [70] Singh, G, Bansal, D. and Sofat. S. 2017. "A smartphone based technique to monitor driving behavior using DTW and crowdsensing." Pervasive and Mobile Computing 40: 56-70. [71] Eren, H.,Makinist, S., Akin, E., and Yilmaz. A., 2012. "Estimating driving behavior by a smartphone." In 2012 IEEE Intelligent Vehicles Symposium, pp. 234-239. IEEE. [72] Johnson, D. A., and Trivedi. M. M. 2011. "Driving style recognition using a smartphone as a sensor platform." 14th International IEEE Conference on Intelligent Transportation Systems (ITSC), pp. 1609-1615. IEEE, 2011. [73] Engelbrecht, J., Booysen, M.J., Rooyen, G.J. and Bruwer. F.J., 2015. "Performance comparison of dynamic time warping (DTW) and a maximum likelihood (ML) classifier in measuring driver behavior with smartphones." In 2015 IEEE Symposium Series on Computational Intelligence, pp. 427-433. IEEE, 2015. [74] Pierobon, M., Marcon, M., Sarti, A., and Tubaro. S. 2005. "Clustering of human actions using invariant body shape descriptor and dynamic time warping." In IEEE Conference on Advanced Video and Signal Based Surveillance., pp. 22-27. IEEE, 2005. [75] Sakakura, Y. 2014. "Household power consumption simulator with compact representation of occupant behaviors." In 2014 IEEE International Conference on Smart Grid Communications (SmartGridComm), pp. 170-175. IEEE, 2014. [76] Wen, L., Zhou, K., and Yang. S. 2019. "A shape-based clustering method for pattern recognition of residential electricity consumption." Journal of cleaner production 212: 475-488. [77] Sarda-Espinosa A. 2019. “dtwclust: Time Series Clustering Along with Optimizations for the Dynamic Time Warping Distance.” R package version 5.5.6, https://CRAN.R- project.org/package=dtwclust. [78] François, P., Ketterlin, A., and Gançarski. P. 2011. "A global averaging method for dynamic time warping, with applications to clustering." Pattern recognition 44, no. 3: 678-693. [79] Shi, K., Qin, H., Sima, C., Li, S., Shen, L., and Ma. O. 2019. "Dynamic barycenter averaging kernel in RBF networks for time series classification." IEEE Access 7: 47564-47576. [80] Cuturi, M., Vert, J.P., Birkenes, O., and Matsui. T. 2007. "A kernel for time series based on global alignments." In 2007 IEEE International Conference on Acoustics, Speech and Signal Processing-ICASSP'07, vol. 2, pp. II-413. IEEE. 137 [81] Dye T.R. 2017. “Understanding Public Policy”, Annual Social and Economic (ASEC) Supplement, 15th Edition, Pearson. pp121-124 [82] Golden, L. 2015. "Irregular work scheduling and its consequences." Economic Policy Institute Briefing Paper 394. [83] Schneider, D, and Harknett K. 2019. "Consequences of routine work-schedule instability for worker health and well-being." American Sociological Review 84, no. 1: 82-114. [84] Enchautegui, M E. 2013. "Nonstandard work schedules and the well-being of low-income families.". [85] Mitra D., Steinmetz N., Chu, Y., Cetin K.S. 2020. "Typical occupancy profiles and behaviors in residential buildings in the United States." Energy and Buildings 210: 109713. [86] Leung, M.Y.M, Groes F., Santaeulalia-Llopis R. 2016. "The relationship between age at first birth and mother's lifetime earnings: Evidence from Danish data." PloS one 11, no. 1: e0146989. [87] Stahl, A. May 1, 2020,10:40am EDT, Forbes, New Study: Millennial Women Are Delaying Having Children Due To Their Careers, https://www.forbes.com/sites/ashleystahl/2020/05/01/new-study-millennial-women-are- delaying-having-children-due-to-their-careers/?sh=41f6c837276a [88] Prototype residential building model, Pacific Northwest National Laboratory (PNNL), https://www.energycodes.gov/prototype-building-models#Residential, Accessed 10th May, 2021 [89] “Guide to determining climate regions by county”, Pacific Northwest National Laboratory, August 2015 [90] EnergyPlus 9.5, https://energyplus.net/, Accessed on 10th May, 2021 [91] Brynjolfsson, E, Horton J.J., Ozimek A., Rock D., Sharma G., and TuYe H-Y. 2020. “COVID-19 and remote work: An early look at US data”. No. w27344. National Bureau of Economic Research. [92] Krantz-Kentkrantz, R. M. 2019. "Where did workers perform their jobs in the early 21st century?." Monthly Labor Review: 1-10. [93] Dunton, G.F., Do B., Wang S.D. 2020. "Early effects of the COVID-19 pandemic on physical activity and sedentary behavior in children living in the US." BMC Public Health 20, no. 1: 1-13. [94] Rundle, A G., Park Y., Herbstman J.B., Kinsey E.W., Wang Y.C. 2020. "COVID‐19–related school closings and risk of weight gain among children." Obesity 28, no. 6: 1008-1009. [95] Moreland, A., Herlihy C., Tynan M.A., Sunshine G., McCord R.F., Hilton C., Poovey J. et al. 2020 "Timing of state and territorial COVID-19 stay-at-home orders and changes in population movement—United States, March 1–May 31, 2020." Morbidity and Mortality Weekly Report 69, no. 35: 1198. 138 [96] Greco, J. February 18, 2021, Cornell Chronicle, Higher-income people take more COVID-19 safety precautions, https://news.cornell.edu/stories/2021/02/higher-income-people-take-more- covid-19-safety-precautions [97] Parker, K, Minkin R., Bennett J. 2020. "Economic fallout from COVID-19 continues to hit lower-income Americans the hardest." Pew Research Center: 21. [98] Kniffin, K M., Narayanan J., Anseel F., Antonakis J., Ashford S.P., Bakker A.B., Bamberger P. et al. 2021. "COVID-19 and the workplace: Implications, issues, and insights for future research and action." American Psychologist 76, no. 1: 63. [99] Boland, B, Smet A.D., Palter R., Sanghvi, A. 2020. "Reimagining the office and work life after COVID-19." McKinsey & Company, June 8. 139 5. CHAPTER 5. COVID-19 IMPACTS ON RESIDENTIAL OCCUPANCY SCHEDULES AND ACTIVITIES IN U.S. HOMES IN 2020 USING ATUS Mitra, Debrudra, Yiyi Chu, and Kristen Cetin. "COVID-19 impacts on residential occupancy schedules and activities in US Homes in 2020 using ATUS." Applied Energy (2022): 119765. https://doi.org/10.1016/j.apenergy.2022.119765 1. Abstract Many aspects of the daily lives of those living in the United States were substantially impacted by the COVID-19 pandemic in the year 2020. A broad diversity of measures was implemented to curb the spread of the virus, many of which included adjustments to where and how people worked, went to school, and otherwise conducted their daily lives compared to pre-pandemic times. This has impacted how residential buildings are used, how much time people spend in their homes, and as a result, how much energy these buildings consume. The main objective of this study is to analyze, at a national scale, the differences in the occupancy schedules and activities conducted in homes in the U.S., as compared to pre-pandemic. 15 years of American Time Use Survey and Current Population Survey data, from 2006 to 2020, was used in this study to analyze the occupancy schedules for both pandemic (2020) and pre-pandemic (2006-2019) times. These impacts were also analyzed with respect to variables including, weekday/weekend, month of the year, age of the occupants, household income, and household size. The impact of the pandemic on occupant schedules were most substantial in the initial months, whereas as the months progressed, these occupancy profiles slowly changed. Across 2020, people spent, on average, 8% more time (1.9 hours) in their home on weekdays, and 3-6% (1.2 hours) on weekend days. The percentage of time spent for different activities and locations within homes were also studied. For 1-member households, their time spent at home decreased whereas for 2-, 3-, and 4- member households, they spent more time at home. Overall, people spent around 45% more time doing office- and work-related activities at home compared to pre-pandemic, which is likely due to increased remote working and schooling. This research helps to improve the understanding of the occupancy presence and absence profiles in U.S. residential buildings due to the pandemic and provides new insights as to modified profiles for researchers, building designers, and policy makers. 140 2. Keywords Occupancy schedules; COVID-19; Residential building; Space utilization; Financial condition 3. Introduction The World Health Organization (WHO) declared COVID-19 to be a global pandemic in March 2020 as the world population faced a new crisis in the form of the spread of the coronavirus [1]. The pandemic has had substantial impacts on peoples’ lives. More than 532 million cases have been reported, with more than 6.3 million deaths worldwide (as of June 2022) [2]. Along with the detrimental effects on human health, the global economy also has faced a 7% loss due to pandemic, as reported in April 2022 [3,4]. This impact on the world economy is expected to be seen for many years; for lower-wealth areas, GDP (gross domestic types) may never fully recover to pre- pandemic levels [5,6]. A significant amount of the workforce lost jobs globally in 2020 compared to what would have been expected without the occurrence of the pandemic [7]. The United Nations also reported that the pandemic resulted in significant reductions in development efforts intended to reduce poverty [8]. The service sector, specifically tourism, was also heavily affected by the pandemic [9, 10]. In summary, there were multidimensional detrimental impacts of the pandemic on all the components of people’s daily life. To curb the spread of the virus, different countries took various measures at varying times. Common approaches included lockdowns and stay-home orders which impacted the way people lived [11-13]. Most K-12 schools and many colleges and universities were shifted to remote learning methods where students learned at home, remotely via computer [14, 15, 16]. In addition, many people also switched to remote work, where they worked from home instead of in an office, representing an estimated 35% of the workforce [6], which was significantly higher than pre-pandemic [17]. A study analyzing phone data in 2020 found that people tended to spend more time in their homes during periods when there were restrictions on travel and gathering, and cancellations of most public events [18]. Although the pandemic forced people to temporarily adapt to a new way of living during these restriction periods, some of the changes in lifestyle have continued well after the measures have been lifted. For example, a global survey of the workforce that worked from home due to COVID [19] found that this shift in working patterns improved workers’ performance in many situations [20]. A survey from October 2020 showed that 54% of employees wanted to continue working 141 from home even after the COVID pandemic whereas pre-pandemic only 20% of the employees surveyed worked remotely [21]. Respondents also preferred to have flexible work schedules, which were more commonly implemented during the work-from-home periods [22-26]. Given these changes, as expected, the use of residential buildings and their corresponding energy consumption behavior has also changed. As occupant behavior is one of the six key parameters which dictates building energy consumption, particularly in residential buildings, it is therefore important to study how COVID has impacted residential occupancy and occupant behaviors [27]. Several studies have discussed the importance and level of uncertainty in energy consumption in residential buildings due to occupant behavior [28-30]. One study found that residential electricity consumption can vary from 0 to 14 kWh/m2 for buildings with similar characteristics in Beijing, China [31]. Plug load and appliance consumption was found to vary by a factor of 10 across homes when occupants have the control over this equipment [32]. Another study showed that the overall energy consumption in similar residential buildings can vary by 3 times due to interaction of occupants with the building systems [33]. Energy consumption patterns also vary significantly based on occupant behavior, particularly in residential buildings [34-38]. Specifically, during the pandemic, various studies have reported utility-level consumption impacts. Overall electricity consumption in Italy had decreased by 28% 1.5 months after the lockdown started [39]. As the number of COVID cases in China reached its peak during the month of March 2020, the maximum reduction in electricity demand was seen during this time [40]. In India, the electricity demand decreased in May 2020 as the number of COVID cases was rising [39]. In the United Kingdom, overall electricity demand decreased by 15% within weeks of the lockdown announcement [41]. Similar trends were also seen in the United States where overall electricity demand fell 19% during the initial stage of the lockdown in April 2020 [42]. The major electric grids in U.S. experienced a reduction in consumption during the first few months of the pandemic, ranging from 3% to 8% for ERCOT (Electric Reliability Council of Texas), PJM (Pennsylvania, New Jersey and Maryland) and MISO (Midcontinent Independent System Operator) [42-44]. The likely reason for reduction in overall consumption is due to the decreased use of commercial and industrial buildings. 142 Different outcomes are seen when comparing the consumption for residential buildings across different countries. A study of 2,000 households in the United Kingdom showed that the electricity consumption increased by 17% during working hours in the initial stage of lockdown [42]. In Ireland, residential energy consumption increased by 11-20% [45]. In Nigeria, the projected share of electricity demand from residential buildings compared to the overall consumption increased from 43% to 49% during lockdown periods in 2020 [46]. In Melbourne, Australia, weather adjusted electricity consumption increased by 14% for residential buildings [47]. Similar trends were seen in the United States. A study in Austin, Texas showed 32% higher electricity use in residential buildings in March 2020 compared to last week of February [48]. This increased consumption in the residential sector can be explained by people spend more time in their homes, and therefore using more electricity for their various activities. This increased use of energy-consuming appliances and devices was also observed. Internet usage increased due to the increased remote working, and escalated use of streaming and social media services [49-51], resulting in increased use. A recent survey showed that during pandemic, people preferred to cook more at home and eat out less, which also would result in more energy consumption [52]. Another study compared heating, ventilation, and air conditioning (HVAC) and non-HVAC consumption in residential building during pandemic and pre-pandemic times showed increased use of weather-normalized HVAC use, as well as increased non-HVAC use [53]. Both of these changes are likely due to changes in occupant behavior, and in occupancy in residential buildings. However, there is limited study of national-level characterization of occupant behavioral changes across U.S. residential buildings as compared to pre-pandemic. Many of the above- mentioned studies include smaller sample sizes and are not representative of the U.S. population as a whole. Studying occupancy patterns in residential buildings and how they differ during the pandemic is therefore needed, to better understand how residential households, overall, have adjusted their occupancy and in-home activities. Such schedules are also important to building energy modeling applications. In current engineering practices, a majority of the building related energy modeling software use existing and predefined schedules based on the Reference Building and Prototype Building models [54,55], ASHRAE 90.1-1989 and the ASHRAE Advanced Energy Design Guide [56, 57], and the Building American Housing Simulation Protocol [60-62]. These schedules were developed pre-pandemic, and some 143 were also based on engineering judgement. However, as it was mentioned in IEA Annex 53, Annex 79 and ISO 18523, occupancy schedules play a significant role in evaluating the energy performance of buildings [27, 58, 59]. While it is recognized that occupancy schedules that result from data collected in 2020 may not be representative of long-term future occupancy post- pandemic, they provide insight into the relative impacts and variations across different segments of the population, as well as overall, based on the most recent available data. In addition, given that trends are more toward remote work in the foreseeable future, studying COVID-19 impacts on residential occupancy provides a preview of potential occupancy trends moving forward, or in the event of another similar type of event [19-26]. The main objective of this study is to evaluate, for the U.S. overall, the typical occupancy schedules of residential building in the United States during the pandemic and compare these results with pre-pandemic occupancy profiles across different timescales and population segments. The results of this analysis can help building designers and energy modelers to understand the pandemic- related changes in occupancy behaviors and schedule patterns in the residential building sector, which might help them to implement modified control and demand-response strategies. This research consists of five sections. Following this introduction, the next section describes the dataset utilized in this research. The third section discusses overall methodology for this study, while the results and discussion are in the fourth section. The final section includes the conclusions and a discussion about limitations of this study and possible future work. 4. Dataset To study the occupancy profiles of residential buildings in the United States, it is beneficial to use a dataset that is representative of the overall population. The American Time Use Survey (ATUS) and Current Population Survey (CPS) are two datasets collected on an annual basis, that are statistically representative of the U.S. population [60, 61]. Both these surveys are administered and managed by the United States Bureau of Labor Statistics and the U.S. Census Bureau and published every year. CPS collects information related to employment, economic, and other characteristics of the U.S. population. Selected households from those who were participated for 8 months or more in the CPS, are selected for participation in the ATUS. Each participant’s data is then weighted using a weighting factor, for use in collectively representing the overall U.S. 144 population. The objective of the ATUS is to collect data on how the U.S. population spends their time, in particular what activities they perform and where they perform them. This survey also is linked with additional data on financial, economic, and social characteristics of the participants. In the ATUS, participants self-report their location and activity information for a single day at 5- minute intervals, over a 24-hour period, from 4 am to 3:59 am the following day. The activities reported are classified into 470 types across 3 major tiers. Only primary activities are stored; secondary activities, i.e., the activities occurred concurrently with the primary activities, are not reported. This survey was first published in the year 2003 and has continued to be publish annually. However, since the method used to calculate the weighting factor was changed in 2006, only data from 2006-2020 (15 years) has been used in this research. It should be noted that the ATUS data collection was, like many efforts, impacted by the pandemic. No data was collected during the start of the lockdown, from mid-March to mid-May 2020. The unavailability of data for these 2 months makes it difficult to evaluate how occupancy profiles were impacted in the initial stage of the pandemic. However, the ATUS data, after mid-May 2020, provides an important source of household activity and occupancy schedules for use in exploring the impact of the pandemic on the daily lives of the U.S. population. 5. Methodology From each year of ATUS data, occupant activity and location information and their characteristics including age, number of household members, day of the week, and month of the year of the collected data were extracted. Similarly, from the CPS, household income information was extracted. Both the datasets have an occupant identifier for use in linking the surveys. After extracting and combining the data from these two datasets, location and activities were mapped to the presence/absence of the participant in a residential building. After the mapping, the presence and absence of occupants from home (residential spaces) was divided into 5-minute timesteps, and translated to a 0 or 1, where 0 represents absence and 1 implies presence within a residential building. It was assumed that the reported location and activity remained constant throughout the 5-minute time intervals between reports. Both the location and activity information were also converted to a schedule from 12:00 am to 11:59 pm in order to be compatible with a typical day, and with the required format of schedules used in energy modeling tools. 145 As the objective of the study is to analyze how occupancy schedules were impacted by various variables including month of the year, day of the week, occupant age, household income, and number of household members, the processed and converted mapped occupancy and location data were subdivided into multiple groups. To do so, first, averaged occupancy profiles were subdivided into weekdays and weekends. These occupancy data were also divided by month across 2018, 2019 and 2020 for comparison. Occupancy information was also divided based on occupant age, including 7 age groups: under 25, 25 to 34, 35 to 44, 45 to 54, 55 to 64, 65 to 74 and over 75, which is consistent with the Residential Energy Consumption Survey (RECS) data [62]. This is driven by findings from other studies that suggest that age groups were likely impacted differently by the pandemic [63, 64, 65]. As occupancy profiles also vary with the number of household members, as shown in previous studies [62, 64], households were also studied based on the number of members. Household income was also evaluated, as recent literature has suggested differential impacts from the pandemic on households across different income levels [63, 66]. This was evaluated by dividing the data into three income groups, low income (LIH), middle income (MIH) and high- income households (HIH). Since the number of household members is important for this calculation, in addition to household income, this is also reported in Table 1 [63, 66]. The thresholds for LIH and HIHs were selected based on recommendations given by U.S. Office of the Assistant Secretary for Planning and Evaluation (ASPE) and previous studies [63, 67, 68]. MIHs included all households with incomes in between these two thresholds. Table 5-1. Threshold household incomes for LIH and HIH income households in 2020 [Note: Total gross incomes are considered here] Number of members 1-person 2-person 3-person 4-person Low Income Household (LIH) < $ 15,000 < $ 20,000 < $ 25,000 < $ 30,000 Threshold High Income Household (HIH) > $ 75,000 > $ 100,000 > $ 125,000 > $ 150,000 Threshold After the calculating average profiles for different household types, more detailed analysis was done on the monthly variation in occupant schedules. To identify the specific type of profiles, cluster analysis was performed across all months on weekdays and weekends. To group the time 146 series profiles, a Dynamic Time Wrapping (DTW) cluster algorithm was used, as the advantage of DTW is that it can group the profiles based on patterns even if they are not time synchronized [69,70]. Several studies have used DTW algorithm to predict occupant behavior and their influence on building consumption [71-73]. To implement the algorithm, the ‘dtwclust’ package was used in R where three clusters were used, recognizing that prior studies have identified 3 major occupancy patterns in ATUS data [74]. Patterns were analyzed and are discussed across each of the clusters. The location distribution of occupants within their home was also evaluated. To do so, all activities defined as being in a residential building were divided into 7 different groups, specifically those located in the bedroom, bathroom, living room, kitchen/dining room, office room, garage, and other areas (e.g. laundry, storeroom). All activities reported in the ATUS were then mapped to all the 7 indoor locations. As an example, sleeping activities were mapped to bedroom, whereas food preparation and kitchen cleaning activities were mapped to the kitchen or dining room. This mapping is done as ATUS data does not the specific location within their home. The percentage of time distribution was then calculated for occupants based on the time when occupant was present in their home. This analysis represents how people spent their time in residential space prior to and during the pandemic. 6. Results and Discussion 6.1. Overall percentage presence in home Percentage presence time (PPT) in residential spaces for people on both weekdays and weekends for the overall population as well as different age groups is shown in Figure 1. These percentage values are compared across the years from 2018 to 2020. The Y-axis represents percentage of the day people spent at home where 100 represents people staying at home for 24 hours whereas 0 represents the absence of people from their home throughout this period. As shown in Figure 1.a, on weekdays, PPT for all the age groups are notably higher in 2020 compared to the two previous years. For those under 25, the average PPT during the pandemic time is 80%, which is 12% higher than in 2018 and 2019. During the pandemic, as most of the school and colleges were switched to remote learning method, students spent most of their time in their home which likely resulted in the higher PPT value. Similarly, as majority of the workplaces switched to work from home, the 147 difference in PPT of people 25 to 54 is around 10%. However, for older people, this value is lower, at 5% and 2% for those 65 to 74 and over 75, respectively. Overall, for all the population, average PPT value increased by around 8% on weekdays (approximately 1.9 hours). A similar trend of higher PPT is also seen on weekends as shown in Figure 1.b. However, unlike weekdays, the difference in the PPT during the pandemic compared to the previous 2 years were almost identical across all age groups, at approximately 3% to 6%. On average, people spent around 1.2 hours more in their homes on a weekend day during the pandemic (in 2020). (a) (b) 100 100 94 91 90 90 88 88 88 87 89 90 89 90 86 87 87 83 85 85 84 83 82 Percentage time Percentage time 83 79 83 83 80 80 80 80 80 79 80 80 80 81 80 76 76 75 73 73 80 73 72 69 68 70 66 66 66 65 65 65 70 60 60 50 50 under 25 25-34 35-44 45-54 55-64 65-74 over 75 overall under 25 25-34 35-44 45-54 55-64 65-74 over 75 overall 2020 2019 2018 2020 2019 2018 Figure 5-1. Percentage presence time distribution for the U.S. (“overall”) and for different age groups for (a) weekdays and (b) weekends for 2018, 2019 and 2020 based on ATUS data 6.2. Variation in occupancy on weekdays and weekends Average profiles for weekdays and weekends prior to and during 2020 are shown in Figure 2 where the y-axis represents the occupancy fraction, i.e., the fraction of people in a home with respect to total number of people in the studied group. The higher the value of the occupancy fraction (OF), the higher the probability of people being in a residential building. An OF 1.0 implies that all the occupants are at home, whereas 0.0 OF represent no one at home. The x-axis represents the time of day, starting at midnight. As shown in Figure 2, for pre-pandemic periods, for both weekday and weekends the OF is high in the early morning, decreases slowly until around noon, then increased gradually to its maximum at night. This also indicates that people stay at home more on weekends compared to weekdays, as to be expected. The change in residential occupancy is substantial compared to the consistency of the prior years. From 2006 to 2019, the calculated occupancy profiles from ATUS data have remained almost identical, with a minimum OF of around 0.4 in weekdays. This increased to approximately 0.55 during the pandemic. Overall distribution of the average profiles is also 148 compared and shown in Figure 3. From Figure 3, both the mean and variance of the OF remained almost constant for past 14 years, from 2006 to 2019. However, in 2020, mean of average OF increased significantly from 0.7 to 0.8 and the variance also reduced. This signifies that, during weekdays of pandemic, people spent more time at home and the variation throughout the day is smaller. Similar results can also be seen on weekends as the difference between the minimum OF was more than 0.1 (Figure 2) along with the higher mean and lower variance (Figure 3). This indicates that more people stayed at home throughout the week during the pandemic. For both weekdays and weekends, the OF reached its peak at an earlier time in 2020, compared to previous years, which signifies that nearly all people were at home for a longer duration during this period. 2020 2019 2018 2017 2016 2015 2014 2013 2012 2011 2010 2009 2008 2007 2006 Figure 5-2. Average daily occupancy profile variation for (a) weekdays and (b) weekends for 2006 to 2020 based on ATUS data Figure 5-3. Mean and variance of average daily occupancy profiles for (a) weekdays and (b) weekends for 2006 to 2020 based on ATUS data 149 6.3. Monthly Variation The variation in the OF across the months of 2020 in comparison to 2019 and 2018 is shown in Figure 4. Due to the unavailability of data for April 2020 in the ATUS, the OF value of April is not shown in this Figure. As shown, for the first three months, from January to March, average OF values were similar across all three years, which is expected given the pandemic began in mid- March and given that prior analysis of ATUS data indicated high levels of consistency in derived occupancy data [64]. 150 Occupancy fraction Occupancy fraction Occupancy fraction 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0:00 0:00 0:00 1:15 1:15 1:15 2:30 2:30 2:30 3:45 3:45 3:45 5:00 5:00 5:00 6:15 6:15 6:15 7:30 7:30 7:30 8:45 8:45 8:45 10:00 10:00 (d) 10:00 (a) 11:15 11:15 11:15 Occupancy fraction 12:30 (g) 12:30 12:30 0.4 0.5 0.6 0.7 0.8 0.9 1.0 13:45 13:45 13:45 15:00 15:00 15:00 0:00 16:15 16:15 16:15 1:15 17:30 17:30 17:30 2:30 18:45 18:45 18:45 3:45 20:00 20:00 20:00 5:00 21:15 21:15 21:15 6:15 Figure 5-4. Average daily occupancy profile for months of (a) January, (b) February, (c) March, 22:30 22:30 22:30 7:30 23:45 23:45 23:45 8:45 Occupancy fraction Occupancy fraction Occupancy fraction Occupancy fraction 10:00 1.0 0.9 0.8 0.7 0.6 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.5 0.4 11:15 (j) 0.4 0.5 0.6 0.7 0.8 0.9 1.0 12:30 0:00 0:00 0:00 0:00 13:45 1:15 1:15 1:15 1:15 15:00 2:30 2:30 2:30 2:30 16:15 3:45 3:45 3:45 3:45 Dec_20 2020 17:30 5:00 5:00 5:00 5:00 18:45 6:15 6:15 6:15 6:15 20:00 7:30 7:30 7:30 7:30 21:15 8:45 8:45 8:45 8:45 22:30 10:00 10:00 (h) 10:00 10:00 151 23:45 11:15 11:15 11:15 11:15 Occupancy fraction 12:30 (e) 12:30 (b) Dec_19 2019 12:30 12:30 13:45 0.4 0.5 0.6 0.7 0.8 (k) 0.9 1.0 13:45 13:45 13:45 15:000:00 15:00 15:00 15:00 16:151:15 16:15 16:15 16:15 17:302:30 17:30 17:30 17:30 18:453:45 18:45 18:45 18:45 Dec_18 20:005:00 20:00 20:00 20:00 2018 21:156:15 21:15 21:15 21:15 22:30 22:30 22:30 (d) May, (e) June, (f) July, (g)August, (h) September, (i) October, (j) November, (k) December 22:307:30 23:458:45 23:45 23:45 23:45 Occupancy fraction Occupancy fraction Occupancy fraction 10:00 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.4 0.5 0.6 0.7 0.8 0.9 1.0 11:15 (k) 12:30 0:00 0:00 0:00 13:45 1:15 1:15 1:15 15:00 2:30 2:30 2:30 16:15 3:45 3:45 3:45 17:30 5:00 5:00 5:00 18:45 6:15 6:15 6:15 20:00 7:30 7:30 7:30 21:15 8:45 8:45 8:45 22:30 10:00 10:00 10:00 23:45 11:15 (i) 11:15 (f) 11:15 (c) 12:30 12:30 12:30 13:45 13:45 13:45 15:00 15:00 15:00 16:15 16:15 16:15 17:30 17:30 17:30 18:45 18:45 18:45 20:00 20:00 20:00 21:15 21:15 21:15 22:30 22:30 22:30 23:45 23:45 23:45 for 2018, 2019 and 2020 based on ATUS data minimum average OF values was around 0.5. However, in the months after the pandemic began, pandemic compared to pre-pandemic years. As shown, for all months before pandemic, the suggests that the average occupancy profile for each month is significantly different during years for each month during and before pandemic, resulting in p-values is less than 2*e-16. This years is noted in Figure 4.d to 4.k, from May to December. A paired t-test was used across the A significant difference is noted in the average OF profile in 2020 as compared to the previous i.e., in May 2020, the monthly average OF values was close to 0.8. The difference in the minimum OF was more than 0.25 for approximately 5 hours. This suggests that people spent significant larger amounts of time at home. In May 2020, the OF value reached its second peak around 8:00 pm which is 2.5 hours earlier than the previous years, which peaked at approximately 10:30 pm. This suggests that if people left their home, they came back home earlier compared to pre- pandemic. A similar trend to May can be seen in June (Figure 4.e). However, the minimum OF value decreased slightly to 0.7 compared the 0.8 observed in April. This suggests that more people began to go out more as time continued, although they continued to return earlier than pre-pandemic. The minimum OF values then gradually decreased from 0.7 to 0.6 between June and September 2020. In addition, during these months, the time in evening that the peak OF values gradually shifted to later. This implies that people slowly began to go out of their homes for longer periods, staying out later in the day. For the last three months of the year, however, this trend leveled off, with the average OF values remaining similar in terms of the OF value and the time of day when those OF values occurred. The minimum OF was close to 0.7, and the evening peak occurred around 9:00 pm, as compared to minimum OFs of 0.6-0.7 and evening peaks at 11:00+ pm pre-pandemic. In summary, the overall occupancy schedules in 2020 varied substantially from that of other years, however these differences between years varied depending on the month of the year, as the pandemic progressed. 6.4. Variation by household size The occupancy schedule variation of households of different sizes (1- to 4-members) is shown in Figure 5 for both weekdays and weekends, representing more than 90% of households in the U.S. [64]. Subfigures in the left column depicted weekday profiles, whereas weekends are on the right. The red dashed line indicates the average OF in 2020, whereas the solid lines represented the profiles for other years. 152 (a1) (a2) 1.0 1.0 0.8 0.8 Occupancy fraction Occupancy fraction 0.6 0.6 0.4 0.4 0.2 0.2 0.0 0.0 0: 1: 2: 3: 4: 5: 6: 7:00 00 00 00 00 00 00 00 0: 1: 2: 3: 4: 5:00 00 00 00 00 00 8: 9: 10 11 12 13 14 15 00 00 :0 :0 :0 :0 :00 0 0 0 0 6: 7: 8: 9: 10 11 12 13 00 00 00 00 :0 :0 :0 :00 0 0 0 16 17 18 19 20 21 22:0 :0 :0 :0 :0 :0 :0 :00 0 0 0 0 0 0 0 14 15 16 17 18 19 20 21:0 :0 :0 :0 :0 :0 :0 :00 0 0 0 0 0 0 23:00 22 23:0 :00 0 0 (b1) (b2) 1.0 1.0 0.8 0.8 Occupancy fraction Occupancy fraction 0.6 0.6 0.4 0.4 0.2 0.2 0.0 0.0 0: 1: 2: 3: 4: 5: 6: 7:00 00 00 00 00 00 00 00 0: 1: 2: 3: 4: 5:00 00 00 00 00 00 8: 9: 10 11 12 13 14 15 00 00 :0 :0 :0 :0 :00 0 0 0 0 6: 7: 8: 9: 10 11 12 13 00 00 00 00 :0 :0 :0 :00 0 0 0 16 17 18 19 20 21 22:0 :0 :0 :0 :0 :0 :0 :00 0 0 0 0 0 0 0 14 15 16 17 18 19 20 21:0 :0 :0 :0 :0 :0 :0 :00 0 0 0 0 0 0 23:00 22 23:0 :00 0 0 (c1) (c2) 1.0 1.0 0.8 0.8 Occupancy fraction Occupancy fraction 0.6 0.6 0.4 0.4 0.2 0.2 0.0 0.0 0: 1: 2:00 00 0: 1: 2: 3: 4: 5: 6: 7:00 00 00 00 00 00 00 00 3: 4: 5: 6: 7: 8: 9:00 00 00 00 00 00 00 00 8: 9: 10 11 12 13 14 15 00 00 :0 :0 :0 :0 :0 :00 0 0 0 0 10 11 12 13 14 15 16 17:0 :0 :0 :0 :0 :0 :0 :00 0 0 0 0 0 0 16 17 18 19 20 21 22 23:0 :0 :0 :0 :0 :0 :00 0 0 0 0 0 0 0 18 19 20 21 22 23:0 :0 :0 :0 :0 :00 0 0 0 0 0 0 :00 (d1) (d2) 1.0 1.0 0.8 0.8 Occupancy fraction Occupancy fraction (d1) 0.6 0.6 1.0 0.8 0.4 Occupancy fraction 0.4 0.6 0.2 0.4 0.2 0.2 0.0 0.0 0.0 0: 1: 00 00 12 12 13 13 104 :0 :3 :0 :3 :00 14 0 0 0 0 0 0 16 17:00 0 2: 3: 4: 5: 00 00 00 00 0 6: :00 000 7: 1 0:3 00:00 8: 1:3 1::30 1 5 00 : 2 00 1 5:0 136: 16 :3 0 :00 4::3 17 178 000 5::00 1 7 00 :3 168:0 :0 00 00 0 0 18 19 20:0 :0 :00 0 0 00 0 2 9: :00 020 10 :30 :03: 11 3 000 :0 :30 12 4:0 :0 0 :05 14 00 0 0 13 4:30 0: 19 9 ::030 8::00 1 9 00 :3 2 0:0 : 0 120 00 0::0 1211: 21 0 12 :30 2 2:00 13 :00 00 3 00 :000 21 22:0 :0 :00 0 0 :05:3 15 6: :0 00 :0 00 0 16 6:30 70 17 :00 :07: 18 8 003 :0 :00 19 8:3 :0 0 :09 0 0 20 9:00 0: 2 2:0 :3 0 1243 0 ::000 1253:3 :00 0 23:00 21 1 30 :00:0 22 10:00 :0 30 10 23 1:00 :1010:3 Year 2020 Year 2019 Year 2018 Year 2017 Year 2016 Year 2015 Year 2014 Year 2013 Year 2012 Year 2011 Year 2010 Year 2009 Year 2008 Year 2007 Year 2006 Figure 5-5. Average occupancy fraction of householder for (a) 1-member, (b) 2-member, (c) 3- member and (d) 4-member households in (1) weekdays and (2) weekends To compare the two samples, a Wilcoxon Signed Rank Test” which is a non-parametric hypothesis test based on the location of population of data was conducted for households with different numbers of members for both weekdays and weekends across the pandemic and pre-pandemic years (Table 2). To analyze the result, the smaller sum of positive and negative ranks needed to be evaluated and compared with respect to the sample size. Corresponding the sample size of 48 and significance level of 0.005, the critical value from Wilcoxon Signed Rank Test is 318. This is higher than the test statistics for all type of households (268 for 1-member household and 0 for others), it represents the profiles differ statistically for households with different member due to the pandemic and thus justify the difference in occupancy pattern during 2020 compared to 153 previous years. As shown in the Figure 5, for both weekdays and weekends, the OF changed notably during the pandemic compared the previous years. In addition, a higher deviation in minimum OF was seen on weekdays compared to weekends. For households of all sizes, the OF value increased throughout the day for both weekdays and weekends and the amount of increase varied with time of day and type of days. For these types of households, the difference increased from early morning and reached its maximum value during the daytime, between 9 am and 3 pm, then slowly decreased until midnight. As such, the occupancy profile was impacted most from 9 am to 3 pm, which would typically, pre-pandemic, be a working period outside of the home. On weekdays, the difference in occupancy profile compared to pre-pandemic was similar for 3- and 4-member households whereas it was slightly smaller for 2-member households. This may be because having a larger household increased the likelihood of the presence of children who require care at home. Specifically for 3- and 4-member households, people also left their home later in morning and returned home earlier in night on both weekdays and weekends. Table 5-2. Non-parametric test of occupancy profiles for different type of households due to pandemic 1-member 2-member 3-member 4-member Sum 268.00 0.00 0.00 0.00 Sample size 48 48 48 48 However, the variation in occupancy profile for 1-member households was different compared to other household sizes. For 1-member households, the OF pattern was similar before and during the pandemic. The main difference was that the profiles were shifted slightly toward the right, suggesting that people left their home later in the morning and returned home later in the evening. On weekends, the first half of the day is very similar, whereas for the second half people stayed out of their home more compared to pre-pandemic. In summary, this analysis suggests that different type of households reacted differently to the pandemic and lockdown measures. 6.5. Variation by occupant age group The ATUS was then divided into 7 age groups, and for each the average of the OF was evaluated in 2020 compared to prior years (Figure 6), including both weekdays and weekends. Figure 6.a shows a comparison of pre-2020 and 2020 average OF for age groups under 45; Figure 6.b shows age groups 45-64, and Figure 6.c shows those over 65. For all age groups, the average OF increased 154 notably in 2020 across all days. Schedules for those under 25 were impacted significantly, as shown in Figure 6.a. For this age group, people were at home more often during pandemic, likely because most schools and colleges were remote. A similar trend is seen for the other age groups. The maximum difference in average OF for people between ages 25 to 64 was around 0.25 for both weekdays and weekends whereas for those younger than 25 and older than 64, it was around 0.3. Also, for all age groups, on average people left their homes later and returned earlier compared to pre-pandemic. Figure 5-6. Variation in occupancy schedule for different age group before and during pandemic 155 6.6. Variation by household income The average OF values comparison across years among LIH, MIH and HIHs for both weekday and weekend are shown in Figure 7. Until 2019, irrespective of weekdays or weekends, the average OF for LIHs is higher than MIHs and HIHs. On weekends, people from LIHs spent around 21 hours at home whereas this value was around 17.5 hours on weekdays. However, for those in HIHs, these values are 19 and 16 hours, respectively, for weekends and weekdays. This suggests that those belonging to HIHs spent the least time at home, whereas those in LIH stayed at home the most compared to other income groups. However, this pattern changed significantly during 2020. During weekdays of 2020, average OF for MIH and HIHs are similar, and they are higher than the OF of LIH for the majority of the day. People in HIHs, and MIHs spent around 20.8 and 19 hours at home in weekends and weekdays respectively whereas these values for LIH are 20.7 and 18.3 hours. This signifies that, people in HIH and MIH likely switched to working remotely and stayed at home more. However, for occupants from LIH, they still needed to leave their home during the pandemic. In addition, interestingly, they spent more time outside their home in 2020 compared to previous years. On weekdays, however, the average OF of LIH is slightly lower midday, from 6 am to 6 pm compared to MIHs and HIHs. This may be because low-income households may include those working in service positions on weekdays that require in-person work without the option for remote work or less likely to switch to remote working [75, 76]. During weekends of 2020, the average OF for all three income groups are similar without much variation between them. Figure 5-7. Variation in occupancy of low (LIH), middle (MIH) and high (HIH) income households on (a) weekdays and (b) weekends during 2020 156 The impact of the 2020 occupancy scenarios on the energy use of a residential building is also evaluated. A single-family prototype residential building is selected as a case study. Weather data for Lansing, Michigan is used, which is a heating dominated climate and located in ASHRAE climate zone 5A. Occupancy-based setback temperature method is used for HVAC control, where when occupant(s) are present, the system runs at the specified setpoint temperature. However, when no occupants are present, the system runs using a 5℃ setback, i.e., heating setback setpoint is 5℃ less than the heating setpoint temperature and similarly, for cooling, the setback setpoint is 5℃ higher than the baseline setpoint temperature. From the average occupancy schedule, when the occupancy fraction is below 50%, the space was assumed to be empty and for other scenarios, occupants are present in the space. EnergyPlus was used to evaluate the building performance and the impact on the HVAC energy consumption is shown in Figure 8 As shown there is negligible variation in consumption for lower income households. However, for MIH and HIH, energy consumption increased significantly during the pandemic. For MIH, overall HVAC consumption increased by around 10%. For HIH, the cooling energy consumption increased by more than 50%. Total HVAC consumption is also increased by 15% for HIHs along with total increase in electricity consumption is around 10%. 60 Percentage change in 50 40 30 20 HVAC consumption (%) 10 0 LIH MIH HIH Heating Cooling Total HVAC Total Gas Total Electricity Figure 5-8. Estimated increase in residential HVAC energy consumption during 2020 compared to prior years in case study location of Lansing, MI (ASHRAE Climate Zone 5A) based on adjusted occupancy profiles 6.7. Indoor location variation The indoor location and primary activity variations of people on both weekdays and weekends were next studied and compared across years as shown in Figure 8, as well as Table 3. Indoor 157 locations include bedroom (BR), bathroom (BT), living room (LR), dining room/kitchen (DR), office/study room (OR), other (OT), and garage (GR); corresponding mapping between the primary activities and the indoor location is given in Appendix A. Figure 5-9. Average time spent in different locations within a home on weekdays for all people in (a) weekdays and (b) weekends for year (1) 2020, (2) 2019 and (3) 2018 (Note: percentages are based on those people reported to be at home, and does not include those outside of the home in the calculation) Table 5-3. Percentage time distribution, by room type, that people spent in their home 2020 2019 2018 Bed (BR) 58.7 60.6 60.8 Bath (BT) 3.0 3.7 3.5 Living (LR) 22.7 22.9 22.8 Dining/kitchen (DR) 7.4 6.9 6.9 Office (OR) 6.2 4.3 4.2 Other (OT) 1.7 1.5 1.6 Garage (GT) 0.2 0.2 0.2 From Table 3, it can be seen that the usage of OR space in 2020 increased by around 45% compared to year 2019. Total time spent in DR space also increased slightly in 2020. However, it is important to note that this percentage distribution is calculated based on the total time people spent at home. Thus, as people spent more time at home in 2020 compared to previous years, the change in total time duration in these spaces are even higher. As shown in Figure 8, the overall distribution of time spent varies in 2020 compared to previous years. On weekdays in 2020, the most significant increase can be seen in OR usage. Pre-2020, on weekdays, OR usage remained consistent, at around 5 – 7% throughout the day (8 am to 8 pm). However, this value increased to around 20% 158 in 2020 from 8 am to 5 pm and then reduced to around 7%. As a significant amount of people worked and went to school remotely, this led to increase in the usage of at-home office spaces. The usage of the dining room/kitchen space also increased slightly to 13% during 12 to 1 pm in 2020, as compared to from 9-10% in prior years. This may be due to people having lunch at home. The use of this space also increased during dinner time to 26% in 2020 from 20% in the previous years, likely due to similar reason. The impact of increased usage of OR and DR space also resulted in reduced usage of the BR space on weekdays. The use pattern of indoor spaces also changed during weekends in 2020 (Figure 8.b). In the morning pre-2020, the living room space was used by around 30% of people, whereas this reduced to 20% during 2020. Usage of the dining room/kitchen space also decreased in 2020. As in previous years, the usage of this space remained consistently uniform with two small spikes during the typical lunch and dinner periods. However, in 2020, the dining room/kitchen space usage decreased in the morning and again in between lunch and dinner time. At the same time, it can be seen that for the office space, percentage distribution of time in weekends is lesser in 2020 compared previous years. As these percentage values are evaluated based on total time people spent in home and during 2020 people spend more time in home compared to 2018 and 2019, total time duration spent in office space is higher during the pandemic. Detailed analysis of space usage for people of different age group is also evaluated for year 2020 and shown in Figures 9 and Appendix B1 for weekdays and weekend respectively. 159 Figure 5-10. Average time spent in different locations within a home on weekdays in 2020 for people (a) under 25, (b) 25-34, (c) 35-44, (d) 45-54, (e) 55-64, f) 65-74 and (g) over 75 (Note: percentages are based on those people reported to be at home, and does not include those outside of the home in the calculation) As shown in Figure 9, starting from midnight to early morning, irrespective of the age group, people spent most of their time in the bedroom, sleeping. In the early morning, the percent of time spent in the bedroom decreases, and other locations increase. For people under 25, the percentage 160 of time spent in their home office is greater in the morning; in the afternoon and evening the living room utilization is greatest. This differs from pre-pandemic (Appendix B2 and B3), where the living room usage dominated throughout the day. This difference may be because of the prevalence of remote learning for schools and colleges, perhaps concentrated more in the mornings and early afternoon. For those age 25 to 64, the time spent in the dining room is comparatively higher in the mornings, likely for breakfast, which then switched to the office space until early evening. The percent time spent in the living room area is low until around 4:00 pm, then increases, likely due to finishing work and transitioning to other activities. Compared to pre-pandemic activity schedules, the usage of living room remains consistent from morning to evening whereas the office space usage is significantly lower for all age groups. Also, unlike 2020, only one spike in dining room usage can be seen during dinner time. For those 65 and older, the time spent in the living room was comparatively higher throughout the day, with the maximum amount for those over 75. Similarly, people in this age range spent significant more time in the dining room/kitchen area, making and eating food, preparing meals, and related activities, during the lunch and dinner time periods. The amount of time spent for meals in the dining/kitchen area is greater for those in older age groups. Compared to pre-pandemic, the major difference is seen in the increased usage of dining/kitchen area spaces. Overall, in 2020, less variation in indoor space usage is seen on weekends, with the primary difference being less time spent in office/study spaces. Those in younger age groups spent more time in the office/study rooms, likely due to work and/or school related activities requiring work on weekends. Overall, the percentage of time spent in the living room, dining room, and bathroom were also higher on weekends compared to weekdays. 7. Conclusions As the COVID-19 pandemic began in the middle of March 2020, it impacted peoples’ lives substantially. This included how people use their homes, the amount of time spent in their homes, and what activities occurred, where, when, and for how long. Previous research has shown that this, as a result, has impacts how buildings use energy and their energy use patterns. This study 161 works towards quantifying what this impact on occupancy and activities occurring in homes has been using a combination of ATUS and CPS data for 2020, compared to the pre-pandemic years of 2006 to 2019. The impact of different time, occupant-related, and household characteristics on occupancy during the pandemic was evaluated in this study and compared pre-pandemic times. For time variables, the variation in weekday and weekend profiles and across months were evaluated. The impact by age group was also analyzed. For household characteristics, the differential impact across different household incomes and household sizes were assessed. Finally, the activities and indoor locations used throughout a typical day were compared on weekdays and weekends, to pre-pandemic times. The major findings of this study are as follows: • During the pandemic in 2020, people spent more time at home. Across this period, the average occupancy fraction on weekdays and weekends was 0.15 higher, and 0.1 higher throughout the majority of the day. Overall, people spent around 1.9 and 1.2 hours more time at home on weekdays and weekends, respectively, which also varies with people’s age. On average, people also came back home earlier on both weekdays and weekends during pandemic. • Considering variations across the different months of 2020, in the beginning of the pandemic (May) a maximum occupancy fraction difference of 0.3 was observed compared to pre-pandemic. This decreased slowly throughout the remainder of the year, to 0.15 in September, then remained approximately 0.2 for the last three months of 2020. • For 2-, 3- and 4- member households, the pandemic resulted in an increase from 17.8 (pre- pandemic) to 19.9 hours of time spent at home on weekdays and 20 (pre-pandemic) to 21.1 hours time spent at homes on weekends. 3- and 4-member households were more impacted. However, 1-person households were comparatively less impacted, where time spent in weekends were similar and on weekdays, they spent on average 32 minutes less in their home. On weekdays, the time at which people most commonly left their home was also delayed slightly. On weekends, people spent more time outside their home in the evening. • Across age groups, although different age groups follow different typical occupancy profiles, the overall impact was similar. On average, the occupancy dropped by 0.07-0.08 across a typical 24-hour day, with a maximum difference in occupancy fraction of 0.25 and 0.3 on weekdays and weekends respectively, typically occurring mid-day. 162 • Across different household incomes, the average occupancy profiles on weekends were similar. However, on weekdays, people in low-income households spent less time at home during the day compared to the other two income groups. Pre-pandemic, people in higher income group spent the least time at home (~16 hours) and people from low-income households spent the maximum (~20 hours). This pattern reversed during the pandemic where people from low-income households spent around 18.6 hours at home which is 0.5 hours less than people from the high income group. • Regarding activities and locations that people spent time in their homes, during the pandemic on weekdays, most time in the morning before noon was spent in office areas, primarily doing remote working and schooling; in the afternoon people continued to work and then transitioned more towards leisure activities, mostly corresponding to the living room area. On weekends, the living room and dining room/kitchen were the two most used spaces across all age groups, whereas younger age groups also used the office/study spaces during this time. Compared to pre-pandemic, the office and kitchen/dining space usage increased significantly whereas the living room and bedroom usage decreased. There are several limitations associated with this study. The ATUS is based on self-reported data and has followed a consistent methodology for data collection since 2006. The ATUS data is not collected specifically to understand COVID-related impacts on occupant behavior. Therefore, direct correlation between COVID and people’s behavior cannot be confirmed. Further research is needed to confirm the specific reasons for changes in people’s behavior in 2020, including but not limited to the impacts of COVID. Also, as the ATUS is self-reported data, it can be subjected to human error. In addition, the survey contains activity data for individual household members for individual, representative days. The availability activity data for multiple days and for multiple household members is not currently available using existing nation-wide survey mechanisms, however this would be helpful for future work to better understand variation in occupancy schedules within households. In addition, some activities reported in the ATUS also include location information while others are assumed based on the nature of the activity. In general, this study provides a detailed analysis of occupancy profiles and how people’s activities and occupancy have been impacted by the pandemic in 2020. The impact of different variables due to the pandemic provides detailed insights that can help in understanding how the pandemic 163 may impact the use of residential buildings long-term, and how these profiles vary across different population segments. For future work, as the availability of ATUS data for 2021 becomes available, it would be highly insightful to conduct further analysis to link what has been found in the 2020 data to understand longer term trends in residential occupancy and how they have varied in the various stages of the pandemic. 8. Acknowledgements The information, data, or work presented herein was funded in part by the Advanced Research Projects Agency-Energy (ARPA-E), U.S. Department of Energy, under Award Number DE- AR0001288. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof. The author also acknowledges the support of the Sloan Foundation. 164 REFERENCES [1] Coronavirus Disease (COVID-19) – Events as they happen., World Health Organization (WHO), https://www.who.int/emergencies/diseases/novel-coronavirus-2019/events-as-they- happen [accessed Sep 29, 2020]. [2] “Coronavirus World Map: Tracking the Global Outbreak”, https://www.nytimes.com/interactive/2021/world/covid-cases.html [3] Recoveries, Managing Divergent. "WORLD ECONOMIC OUTLOOK." (2021). [4] Yeyati, E. L., Filippini, F., 2021. “Social and economic impact of COVID-19”, June 8, 2021, https://www.brookings.edu/research/social-and-economic-impact-of-covid-19/ [5] Djankov, S, Panizza, U. 2020. “COVID-19 in developing economies.” Centre for Economic Policy Research. [6] “Effects of the coronavirus COVID-19 pandemic (CPS)”, U.S. Bureau of Labor Statistics. Accessed: Sep. 26, 2020. [Online]. Available: https://www.bls.gov/cps/effects-of-the- coronavirus-covid-19-pandemic.htm#. [7] Monitor, I. L. O. 2020. "COVID-19 and the world of work. Seventh edition, 2020. " Updated estimates and analysis”. [8] United Nations, 2021. “COVID crisis to push global unemployment over 200 million mark in 2022”, June 2nd, 2021 [9] Krarti, M, Aldubyan, M. 2021. "Review analysis of COVID-19 impact on electricity demand for residential buildings." Renewable and Sustainable Energy Reviews: 110888. [10] UNWTO. 2020. “International tourism numbers could fall 60-80%. Madrid, Spain: United Nations World Tourism Organization”. https://www.unwto.org/news/covid-19-international- tourist-numbers-could-fall-60-80-in-2020. [11] Gibney, E. 2020. "Whose coronavirus strategy worked best? Scientists hunt most effective policies." Nature 581, no. 7806: 15-17. [12] Nicola, M., Alsafi, Z., Sohrabi, C., Kerwan, A., Al-Jabir, A., Iosifidis, C., Agha, M., Agha, R. 2020. "The socio-economic implications of the coronavirus pandemic (COVID-19): A review." International journal of surgery 78: 185-193. [13] “See which states and Cities have Told Residents to stay at home”, The New York Times. Accessed: Sep. 19, 2020. [Online]. Available: https://www.nytimes.com/interactive/2020/us/coronavirus-stay-at-home-order.html. [14] Donohue, J.M., Miller, E., 2020. “COVID-19 and school closures”, American Medical Association, JAMA - Jama. 324(9), pp 845-847. 165 [15] Dunton, G.F., Do B., Wang S.D. 2020. "Early effects of the COVID-19 pandemic on physical activity and sedentary behavior in children living in the US." BMC Public Health 20, no. 1: 1-13. [16] Rundle, A G., Park Y., Herbstman J.B., Kinsey E.W., Wang Y.C. 2020. "COVID‐19–related school closings and risk of weight gain among children." Obesity 28, no. 6: 1008-1009. [17] Krantz-Kentkrantz, R. M. 2019. "Where did workers perform their jobs in the early 21st century?." Monthly Labor Review: 1-10. [18] Moreland, A., Herlihy C., Tynan M.A., Sunshine G., McCord R.F., Hilton C., Poovey J. et al. 2020. "Timing of state and territorial COVID-19 stay-at-home orders and changes in population movement— United States, March 1–May 31, 2020." Morbidity and Mortality Weekly Report 69, no. 35: 1198. [19] “The never normal. Defining new ways of working beyond COVID-19”, Accenture, October 26, 2020. [20] Lee, J., 2020. “Five ways to imagine a better future of work in the post-covid era”, Forbes Human Resources Council, December 14, 2020 [21] Parker, K., Horowitz, J. M., Minkin, R. 2020. “How the Coronavirus Outbreak Has – and Hasn’t – Changed the Way Americans Work, Social & Demographic Trends”, Pew Research Center, 2020. Available, https://www.pewresearch.org/social-trends/2020/12/09/how-the- coronavirus-outbreak-has-and-hasnt-changed-the-wayamericans-work/. [22] Louie, D., 2021. “How COVID-19 pandemic forever changed the workplace, what companies are doing going forward”, ABC7NEWS, March 17, 2021 [23]. Weekes, W., 2020. “13 ways the coronavirus pandemic could forever change the way we work”, CNBC, April 29, 2020. [24]. Mudditt, J., 2020. “How offices will change after coronavirus”, BBC, May 14, 2020 [25]. Davenport, T. H., Redman, T. C., 2021. “Experiments and data for post-COVID-19 work arrangements”, MITSloan Management Review, March 23, 2021 [26]. “How the COVID-19 pandemic is changing the way we work”, Corehealth Technologies, Jul 16, 2020 [27] Yoshino H., Hong T., Nord N. 2017. “IEA EBC annex 53: total energy use in buildings— analysis and evaluation methods”, Energy Build 152, 124–136, doi:10.1016/j.enbuild.2017.07.038 [28] Yan D., Brien W.O., Hong T., Feng X., Gunay H.B., Tahmasebi F. 2015. “Occupant behavior modeling for building performance simulation: Current state and future challenges”, 107, 264– 278. doi:10.1016/j.enbuild.2015.08.032. 166 [29] Yu Z., Fung B.C.M., Haghighat F., Yoshino H., Morofsky E., 2011. “A systematic procedure to study the influence of occupant behavior on building energy consumption”. Energy Build, 43, pp. 1409-1417, 10.1016/j.enbuild.2011.02.002. [30] Lo L.J., Novoselac A., 2010. “Localized air-687 conditioning with occupancy control in an open office”. Energy Build., 42 (7), pp. 1120-1128. [31] Li, Z.-J., Jiang, Y., Wei, Q.-P., 2014, “Survey on energy consumption of air conditioning in summer in a residential building in Beijing”, Heat. Vent. Air Cond., 37 (2007), pp. 46-51. [32] Cetin, K.S., Tabares-Valasco, P.C., and Novoselac, A. 2014. “Appliance daily energy use in new residential buildings: Use profiles and variation in time-of-use”. Energy and Buildings 84(2014): 716-26. [33] Fabi, V., Andersen, R.V., Corgnati, S.P., Olesen, B.W., 2013. “A methodology for modelling energy-related human behaviour: application to window opening behaviour in residential buildings”, Build Simul J (2013), 10.1007/s12273-013-0119-6 [34] An J, Yan D, Hong T. 2018. “Clustering and statistical analyses of air-conditioning intensity 457 and use patterns in residential buildings[J]”. Energy and Buildings, 174: 214-227. [35] Xia, D., Lou, S., Huang, Y., Zhao, Y., Li, D. H., & Zhou, X. 2019. “A study on occupant behavior related to air-conditioning usage in residential buildings”. Energy and Buildings, 203, 109446. [36] Santin, O. G., Itard, L., Visscher, H., 2009. “The effect of occupancy and building characteristics on energy use for space and water heating in Dutch residential stock”, Energy Build, 41, pp. 1223-1232, 10.1016/j.enbuild.2009.07.002. [37] Wilke, U. 2013, “Probabilistic Bottom-up Modelling of Occupancy and Activities to Predict Electricity Demand in Residential Buildings”. PhD Thesis. 5673 172, 10.5075/epfl-thesis-5673. [38] Popoola, O. M. 2018. “Computational Intelligence Modelling Based on Variables Interlinked with Behavioral Tendencies for Energy Usage Profile – A Necessity.” Renewable and Sustainable Energy Reviews 82: 60–72. https://doi.org/10.1016/j.rser.2017.09.020. [39] IEA. Covid-19 impact of electricity – statistical report may 2020. International Energy Agency; 2020. https://www.iea.org/reports/covid-19-impact-on-electricitys. [40] Sambasivam B. 2020. “World’s largest lockdown and its consequences for the electricity sector.” Opinion Editorial. India Times - Energy World; 2020. https://energy.economictimes.indiatimes.com/energy-speak/world-s-largest-lockdown-and-its- consequences-for-the-electricity-sector/4155. [41] Gausden G. 2020. “Home electricity consumption now peaks at 1pm while smart meter data suggests people are having early nights AND late mornings”. Article in This is Money; 2020. https://www.thisismoney.co.uk/money/bills/article-8224837/Energy-usage-risen-17-lockdown- reveal-tips-bills-down.html. 167 [42] Opheim, C. 2020. “COVID-19 Load Impact Analysis”, ERCOT. Available, http://www.ercot.com/content/wcm/lists/200201/ERCOT_COVID-19_Analysis_FINAL.pdf. [43] Gledhill, A. 2020. “Update of COVID-19 Load Impacts”, PJM. Available, https://pjm.com//media/committees-groups/committees/pc/2020/20200512/20200512-item-16- covid-19-load-impact-update.ashx. [44] COVID-19 Impact to LOAD & Outage Coordination, MISO, 2020. Available, https://cdn.misoenergy.org/COVID%2019%20Impacts%20to%20MISO%20Load%20and%20O utage_as%20of%20April%2010%20Final442224.pdf. [45] Savills. 2020. “COVID-19 restrictions changing the daily patterns of energy consumption”. https://www.savills.us/insight-and-opinion/savills-news/299070/covid-19-restrictions-changing- the-daily-patterns-of-energy-consumption. [46] Edomah N, Ndulue G. 2020. “Energy transition in a lockdown: an analysis of the impact of COVID 19 on changes in electricity demand in Lagos Nigeria”. Glob Transit 2020;2 (January):127–37. https://doi.org/10.1016/j.glt.2020.07.002. [47] Farrow H. 2020. “Commercial down v residential up: COVID-19’s electricity impact”. Energy Networks Australia. https://www.energynetworks.com.au/news/energy-insider/2020- energy-insider/commercial-down-v-residential-up-covid-19selectricity-impact/. [48] Oldman I. 2020. “DATA: Austin’s residential electricity usage up more than 30% since beginning of March”, Community Impact Newspaper. https://communityimpact.com/austin/northwest-austin/coronavirus/2020/04/02/data-austins- residential-electricity-usage-up-more-than-30-since-beginning-of-march/. [49] Cohen J. 2020. “Data Usage has Increased 47 Percent during COVID-19 Quarantine”, PCMag. Available, https://www.pcmag.com/news/data-usage-has-increased-47-percent-during- covid-19-quarantine. [50] Yuan, E.S. 2021. “A Year Later: reflecting and Looking Ahead, Blog, Zoom”. Available, https://blog.zoom.us/reflecting-looking-ahead/. [51] Auxier, B., Anderson, M. 2021. “Social media Use in 2021, Internet & Technology”, Pew Research Center. Available, https://www.pewresearch.org/internet/2021/04/07/social-media-use- in-2021/. [52] Renner, B., Cook, J., Rogers, S. 2021 “Surprise Ingredients in the Post-pandemic Food Story: consumers Cooking up Concern for Restaurants’ Return, Deloitte”. Available, https://www2.deloitte.com/us/en/pages/consumer-business/articlesc/food-service-restaurant- business-trends-post-covid.html. [53] Kawka, E., Cetin, K. 2021. "Impacts of COVID-19 on residential building energy use and performance." Building and Environment 205: 108200. 168 [54] Deru, M., Field K., Studer D., Benne K., Griffith B., Torcellini P., Liu B., et al. 2011. “U.S. Department of Energy commercial reference building models of the national building stock.” Nrel/Tp-5500-46861, no. February 2011: 1–118. [55] Goel, R., Athalye, X., Wang, H., Zhang, M. 2014. “Enhancements to ASHRAE standard 90.1 prototype building models”. No. PNNL-23269. Pacific Northwest National Lab (PNNL), Richland, WA (United States). [56] ASHRAE 90.1 User's Manual ANSI/ASHRAE/IESNA Standard 90.1-1989, American Society of Heating, Refrigerating and Air-Conditioning Engineers, Atlanta, GA (1989) [57] Advanced Energy Design Guides, 2018. 724 (accessed October 10, 2020) [58] Annex 79: ANNEX, EBC. "Occupant-centric Building Design and Operation." (2019). [59] ISO 18523-2, “Energy performance of buildings — Schedule and condition of building, zone and space usage for energy calculation — Part 2: Residential buildings” [60] U.S. Bureau of Labor Statistics, American Time Use Survey, (2009). http:// www.bls.gov/tus/#data (Accessed on 10th August 2021.). [61] U.S. Bureau of Labor Statistics, Current Population Survey, https://www.bls.gov/cps/ (Accessed on 10th Sep, 2021.) [62] U.S. Energy Information Administration, Residential Energy Consumption Survey, (2015). [63] Mitra, D, Chu, Y, Cetin, K, Wang, Y, Chen C.F., 2021, “Variation in residential occupancy profiles in the United States by household income level and characteristics”. Journal of Building Performance Simulation, 14:6, 692-711, DOI: 10.1080/19401493.2021.2001572. [64] Mitra, D., Steinmetz, N., Chu, Y., & Cetin, K. S. 2020. “Typical occupancy profiles and behaviors in residential buildings in the United States”. Energy and Buildings, 210, 109713. https://doi.org/10.1016/j.enbuild.2019.109713. [65] Martins Van Jaarsveld, G., 2020. “The Effects of COVID-19 Among the Elderly Population: A Case for Closing the Digital Divide”. Frontiers in Psychiatry, 11. https://doi.org/10.3389/fpsyt.2020.577427. [66] Mitra, D., Yiyi Chu;, & Cetin, K. 2021. Characteristics of Residential Occupancy Profiles for Different Income Groups in the United States. ASHRAE Transactions, v127, 91–99. [67] Office of ASPE, U.S. Department of Health & Human Services, 2018. Poverty Guidelines. https://aspe.hhs.gov/poverty-guidelines. [68] Kochhar, R., Fry R., and Rohal M. 2016. "America’s Shrinking Middle Class: A Close Look at Changes Within Metropolitan Areas." Pew Research Center. 169 [69] Berndt, D.J., and Clifford. J, 1994. "Using dynamic time warping to find patterns in time series." In KDD workshop, vol. 10, no. 16, pp. 359-370. [70] Lawrence, R., Rosenberg, A., and Levinson. S. 1978. "Considerations in dynamic time warping algorithms for discrete word recognition." IEEE Transactions on Acoustics, Speech, and Signal Processing 26, no. 6: 575-582. [71] Singh, G, Bansal, D. and Sofat. S. 2017. "A smartphone based technique to monitor driving behavior using DTW and crowdsensing." Pervasive and Mobile Computing 40: 56-70. [72] Engelbrecht, J., Booysen, M.J., Rooyen, G.J. and Bruwer. F.J., 2015. "Performance comparison of dynamic time warping (DTW) and a maximum likelihood (ML) classifier in measuring driver behavior with smartphones." In 2015 IEEE Symposium Series on Computational Intelligence, pp. 427-433. IEEE, 2015. [73] Sakakura, Y. 2014. "Household power consumption simulator with compact representation of occupant behaviors." In 2014 IEEE International Conference on Smart Grid Communications (SmartGridComm), pp. 170-175. IEEE, 2014. [74] Sarda-Espinosa A. 2019. “dtwclust: Time Series Clustering Along with Optimizations for the Dynamic Time Warping Distance.” R package version 5.5.6, https://CRAN.R project.org/package=dtwclust. [75] Helppie-McFall, B., & Hsu, J. W. 2020. Financial Profiles of Workers Most Vulnerable to Coronavirus-Related Earnings Loss in the Spring of 2020. Finance and Economics Discussion Series, 2020(093), 1–27. https://doi.org/10.17016/FEDS.2020.093 [76] McNicholas, C., and Poydock, M. 2020. "Who are essential workers? A comprehensive look at their wages, demographics, and unionization rates." Economic Policy Institute 19. 170 APPENDIX Table A8. Mapping between indoor locations and primary activities from ATUS TRCODE Activity Description TRCODE Activity Description BED LIVING 10101 Sleeping 10501 Personal emergencies 10102 Sleeplessness 10599 Personal emergencies 10199 Sleeping * 20101 Interior cleaning 10401 Personal/Private activities 20103 Sewing, repairing, & maintaining textiles 10499 Personal activities, n.e.c.* 20199 Housework, n.e.c.* 30101 Physical care for hh children 20301 Interior arrangement, decoration, & repairs 30102 Reading to/with hh children 20302 Building and repairing furniture 30103 Playing with hh children, not sports 20303 Heating and cooling Interior maintenance, repair, & decoration, 30104 Arts and crafts with hh children 20399 n.e.c.* 30106 Talking with/listening to hh children 20601 Care for animals and pets (not veterinary care) 30199 Caring for & helping hh children, n.e.c.* 20602 Walking / exercising / playing with animals 30301 Providing medical care to hh children 20699 Pet and animal care, n.e.c.* 30399 Activities related to hh child's health, n.e.c.* 20905 Home security 30401 Physical care for hh adults 20999 Household management, n.e.c.* 30402 Looking after hh adult (as a primary activity) 29999 Household activities, n.e.c.* 30403 Providing medical care to hh adult 30105 Playing sports with hh children 30499 Caring for household adults, n.e.c.* 30109 Looking after hh children (as a primary activity) 30599 Helping household adults, n.e.c.* 30111 Waiting for/with hh children 40101 Physical care for nonhh children 30203 Home schooling of hh children 40102 Reading to/with nonhh children 30204 Waiting associated with hh children's education 40103 Playing with nonhh children, not sports 30303 Waiting associated with hh children's health Waiting associated with caring for household 40104 Arts and crafts with nonhh children 30405 adults 40106 Talking with/listening to nonhh children 30504 Waiting associated with helping hh adults Looking after nonhh children (as primary 40199 Caring for and helping nonhh children, n.e.c.* 40109 activity) 40301 Providing medical care to nonhh children 40111 Waiting for/with nonhh children Activities related to nonhh child's health, 40399 40203 Home schooling of nonhh children n.e.c.* 40401 Physical care for nonhh adults 40405 Waiting associated with caring for nonhh adults Looking after nonhh adult (as a primary 40402 40503 Animal & pet care assistance for nonhh adults activity) 40403 Providing medical care to nonhh adult 40508 Waiting associated with helping nonhh adults 40499 Caring for nonhh adults, n.e.c.* 50104 Waiting associated with working 40599 Helping nonhh adults, n.e.c.* 50201 Socializing, relaxing, and leisure as part of job 49999 Caring for & helping nonhh members, n.e.c.* 50301 Income-generating hobbies, crafts, and food 171 Table A8 (cont’d) Waiting associated with other income- 80402 Using in-home health and care services 50305 generating activities 80499 Using medical services, n.e.c.* 80403 Waiting associated with medical services 90103 Using clothing repair and cleaning services 80502 Waiting associated w/personal care services 120301 Relaxing, thinking 90101 Using interior cleaning services Waiting associated with using household 120302 Tobacco and drug use 90104 services 120312 Reading for personal interest 90199 Using household services, n.e.c.* Using home maint/repair/décor/construction 120313 Writing for personal interest 90201 svcs Waiting associated w/ home 130109 Dancing 90202 main/repair/décor/constr Using home maint/repair/décor/constr services, 150103 Reading 90299 n.e.c.* 150105 Writing 90301 Using pet services 150203 Providing care 90302 Waiting asso w pet services, n.e.c.* BATH 90399 Using pet services, n.e.c.* Waiting associated with using lawn & garden 10201 Washing, dressing and grooming oneself 90402 services Waiting associated with vehicle main. or repair 10299 Grooming, n.e.c.* 90502 svcs 10301 Health-related self care 99999 Using household services, n.e.c.* 10399 Self care, n.e.c.* 100101 Using police and fire services 80501 Using personal care services 100102 Using social services Waiting associated with using government 80599 Using personal care services, n.e.c.* 100304 services Waiting associated with civic obligations & DINING/KITCHEN 100305 participation 20104 Storing interior hh items, inc. food 120101 Socializing and communicating with others 20201 Food and drink preparation 120199 Socializing and communicating, n.e.c.* Attending or hosting 20202 Food presentation 120201 parties/receptions/ceremonies 20203 Kitchen and food clean-up 120299 Attending/hosting social events, n.e.c.* Food & drink prep, presentation, & clean-up, 20299 120303 Television and movies (not religious) n.e.c.* Housework, cooking, & shopping assistance 40501 120304 Television (religious) for nonhh adults 50202 Eating and drinking as part of job 120305 Listening to the radio 90102 Using meal preparation services 120306 Listening to/playing music (not radio) 110101 Eating and drinking 120307 Playing games 110199 Eating and drinking 120311 Hobbies, except arts & crafts and collecting 110201 Waiting associated w/eating & drinking 120399 Relaxing and leisure, n.e.c.* Waiting associated with eating & drinking, 110299 120501 Waiting assoc. w/socializing & communicating n.e.c.* 119999 Eating and drinking 120502 Waiting assoc. w/attending/hosting social events 150201 Food preparation, presentation, clean-up 120503 Waiting associated with relaxing/leisure OFFICE 120599 Waiting assoc. w/socializing & communicating 20901 Financial management 129999 Socializing, relaxing, and leisure, n.e.c.* 172 Table A8 (cont’d) Household & personal organization and 20902 130105 Playing billiards planning HH & personal mail & messages (except e- 20903 130201 Watching baseball mail) 20904 HH & personal e-mail and messages 130202 Watching baseball 30108 Organization & planning for hh children 130203 Watching basketball 30201 Homework (hh children) 130204 Watching baseball Activities related to hh child's education, 30299 130205 Watching baseball n.e.c.* 30302 Obtaining medical care for hh children 130206 Watching baseball Obtaining medical and care services for hh 30404 130207 Watching bowling adult 30501 Helping hh adults 130208 Watching baseball 30502 Organization & planning for hh adults 130209 Watching dancing 40108 Organization & planning for nonhh children 130210 Watching equestrian sports 40201 Homework (nonhh children) 130211 Watching equestrian sports 40204 Waiting related to nonhh child's educ., n.e.c.* 130212 Watching equestrian sports 40299 Activities related to nonhh child's educ., n.e.c.* 130213 Watching football 40302 obtain medical care for nonhh children 130214 Watching golfing 40303 Waiting related to nonhh child's health 130215 Watching gymnastics Obtaining medical and care services for nonhh 40404 130216 Watching hockey adult Financial management assistance for nonhh 40505 130217 Watching martial arts adults Household management & paperwork 40506 130218 Watching racquet sports assistance for nonhh adults 50101 Work, main job 130219 Watching rodeo competitions 50102 Work, other job(s) 130220 Watching running 50103 Security procedures related to work 130221 Watching running 50199 Working, n.e.c.* 130222 Watching running 50299 Work-related activities, n.e.c.* 130223 Watching skiing, ice skating, snowboarding 50302 Income-generating services 130224 Watching soccer 50303 Income-generating services 130225 Watching softball 50304 Income-generating rental property activities 130226 Watching running 50399 Other income-generating activities, n.e.c.* 130227 Watching volleyball 50401 Job search activities 130228 Watching running 50403 Job interviewing 130229 Watching water sports 50404 Waiting associated with job search or interview 130230 Watching wrestling Security procedures rel. to job 50405 130231 Watching wrestling search/interviewing 59999 Work and work-related activities, n.e.c.* 130232 Watching wrestling Taking class for degree, certification, or Waiting associated w/religious & spiritual 60101 140103 licensure activities 60102 Taking class for personal interest 140105 Religious education activities 60103 Waiting associated with taking classes 150102 Organizing and preparing 173 Table A8 (cont’d) 60199 Taking class, n.e.c.* 150104 Telephone calls (except hotline counseling) 60201 Extracurricular club activities 150202 Collecting & delivering clothing & other goods Waiting associated with extracurricular 60204 150204 Teaching, leading, counseling, mentoring activities Education-related extracurricular activities, Indoor & outdoor maintenance, repair, & clean- 60299 150302 n.e.c.* up Research/homework for class for degree, Indoor & outdoor maintenance, building & 60301 150399 certification, or licensure clean-up activities, n.e.c.* 60302 Research/homework for class for pers. interest 150401 Performing Participating in performance & cultural 60399 Research/homework n.e.c.* 150499 activities, n.e.c.* Administrative activities: class for degree, 60401 150701 Waiting associated with volunteer activities certification, or licensure Waiting associated with volunteer activities, 60402 Administrative activities: class for personal 150799 n.e.c.* 60403 Waiting associates w admin activities 150801 Waiting associated with volunteer activities 60499 Administrative for education, n.e.c.* 150899 Waiting associated with volunteer activities 69999 Education, n.e.c.* 159999 Volunteer activities, n.e.c.* 70104 Shopping, except groceries, food and gas 160101 Telephone calls to/from family members Telephone calls to/from friends, neighbors, or 70105 Waiting associated with shopping 160102 acquaintances Telephone calls to/from education services 70199 Shopping, except groceries, food and gas 160103 providers 70201 Comparison shopping 160104 Telephone calls to/from salespeople Telephone calls to/from professional or personal 70299 Comparison shopping 160105 care svcs providers Telephone calls to/from household services 70301 Security procedure related to purchase 160106 providers Telephone calls to/from paid child or adult care 70399 Security procedure related to purchase 160107 providers 80101 Using paid childcare services 160108 Telephone calls to/from government officials 80102 Paid childcare 160199 Telephone calls (to or from), n.e.c.* 80199 use paid childcare service 160201 Waiting associated with telephone calls 80201 Banking 169999 Telephone calls, n.e.c.* 80202 Using other financial services OTHER Waiting associated w/banking/financial 80203 20102 Laundry services Appliance, tool, and toy set-up, repair, & 80299 Using other financial services 20801 maintenance (by self) 80301 Using legal services 20899 Appliances and tools, n.e.c.* 80302 Waiting associated with legal services 50203 Sports and exercise as part of job 80399 Using legal services 50204 Security precedure as part of job 80601 Activities rel. to purchasing/selling real estate 60104 Security proc rel to take class Waiting associated w/purchasing/selling real 80602 60202 Extracurricular music & performance activities estate 80699 Activities rel. to purchasing/selling real estate 60203 Extracurricular student govt activities 80701 Using veterinary services 120309 Arts and crafts as a hobby 80702 Waiting associated with veterinary services 120310 Collecting as a hobby 174 Table A8 (cont’d) 80799 Using veterinary services 130101 Doing aerobics 80801 Security procedure related to service 130104 Biking 80899 Security procedure related to service 130124 Running 89999 Professional and personal services, n.e.c.* 130128 Using cardiovascular equipment 100103 Obtaining licenses & paying fines, fees, taxes 130131 Walking 100199 Using government services, n.e.c.* 130133 Weightlifting/strength training 100401 Security procedure related to civic obligation 130134 Working out, unspecified 100499 Security procedure related to civic obligation 130136 Doing yoga 109999 Government services, n.e.c.* 130199 Playing sports n.e.c.* 120308 Computer use for leisure (exc. Games) 140102 Participation in religious practices 120405 Security related to art 149999 Religious and spiritual activities, n.e.c.* 140104 Security related to religious activities GARAGE 150101 Computer use 20701 Vehicle repair and maintenance (by self) 150106 Fundraising 20799 Vehicles, n.e.c.* Vehicle & appliance maintenance/repair 150199 Administrative & support activities, n.e.c.* 40504 assistance for nonhh adults 150299 Social service & care activities, n.e.c.* 90501 Using vehicle maintenance or repair services 150501 Attending meetings, conferences, & training 90599 Using vehicle maint. & repair svcs, n.e.c.* Attending meetings, conferences, & training, 150599 n.e.c.* 175 Figure B7. Average time spent in different locations within a home on weekends in 2020 for people (a) under 25, (b) 25-34, (c) 35-44, (d) 45-54, (e) 55-64, (f) 65-74 and (g) over 75 (Note: percentages are based on those people reported to be at home, and does not include those outside of the home in the calculation) 176 100% 90% 100% 90% 80% 100% 90% 80% 80% 70% 60% 70% 60% 50% 70% 60% 50% 50% 40% 30% 40% 30% 20% 40% 30% 20% 20% 10% 0% 10% 0% 10% 0% 0:00 0:00 0:00 0:50 0:50 0:50 1:40 1:40 1:40 2:30 2:30 2:30 3:20 bed 3:20 bed 3:20 bed 4:10 4:10 4:10 5:00 5:00 5:00 bath 5:50 bath 5:50 bath 5:50 6:40 6:40 6:40 7:30 7:30 7:30 living 8:20 living 8:20 living 8:20 9:10 9:10 9:10 10:00 10:00 10:00 10:50 10:50 10:50 dining dining dining 100% 90% 80% 11:40 11:40 35-44 11:40 70% 55-64 60% 50% 40% 12:30 12:30 30% 12:30 Under 25 20% 10% 0% 13:20 13:20 13:20 0:00 office 14:10 office 14:10 office 14:10 0:50 15:00 15:00 15:00 1:40 15:50 15:50 15:50 2:30 other other 16:40 other 16:40 16:40 bed 3:20 17:30 17:30 17:30 4:10 18:20 18:20 18:20 5:00 garage 19:10 garage 19:10 garage 19:10 bath 5:50 20:00 20:00 20:00 6:40 20:50 20:50 20:50 7:30 21:40 21:40 21:40 living 8:20 22:30 22:30 22:30 9:10 23:20 23:20 23:20 Figure B8. Average time spent in different locations within a home on weekdays in 2019 for 10:00 10:50 100% 100% 177 dining 90% 80% 70% 90% 80% 70% 100% 90% 80% 11:40 60% 50% 40% 60% 50% 40% 70% 60% 50% Over 75 30% 20% 10% 30% 20% 10% 40% 30% 20% 12:30 0% 0% 10% 0% 13:20 0:00 0:00 0:00 office 14:10 0:50 0:50 0:50 15:00 1:40 1:40 1:40 15:50 2:30 2:30 2:30 other 16:40 bed 3:20 bed 3:20 3:20 17:30 bed 4:10 4:10 4:10 18:20 5:00 5:00 5:00 garage 19:10 bath 5:50 bath 5:50 bath 5:50 20:00 6:40 6:40 6:40 20:50 7:30 7:30 7:30 21:40 living 8:20 living 8:20 living 8:20 22:30 9:10 9:10 9:10 23:20 10:00 10:00 10:00 10:50 10:50 10:50 dining dining dining 11:40 65-74 11:40 45-54 11:40 25-34 12:30 12:30 12:30 13:20 13:20 13:20 office 14:10 office 14:10 office 14:10 15:00 15:00 15:00 15:50 15:50 15:50 other 16:40 other 16:40 other 16:40 17:30 17:30 17:30 18:20 18:20 18:20 garage 19:10 garage 19:10 garage 19:10 20:00 20:00 20:00 20:50 20:50 20:50 21:40 21:40 21:40 22:30 22:30 22:30 people of different age group 23:20 23:20 23:20 100% 90% 80% 100% 90% 80% 100% 90% 80% 70% 60% 50% 70% 60% 50% 70% 60% 50% 40% 30% 20% 40% 30% 20% 40% 30% 20% 10% 0% 10% 0% 10% 0% 0:00 0:00 0:00 0:50 0:50 0:50 1:40 1:40 1:40 2:30 2:30 2:30 bed 3:20 bed 3:20 bed 3:20 4:10 4:10 4:10 5:00 5:00 5:00 bath 5:50 bath 5:50 bath 5:50 6:40 6:40 6:40 7:30 7:30 7:30 living 8:20 living 8:20 living 8:20 9:10 9:10 9:10 10:00 10:00 10:00 10:50 10:50 10:50 100% 90% 80% dining 11:40 dining 11:40 dining 11:40 70% 60% 50% 55-64 35-44 40% 30% 20% 12:30 12:30 12:30 10% Under 25 0% 13:20 13:20 13:20 0:00 office office office 14:10 14:10 14:10 0:50 15:00 15:00 15:00 1:40 15:50 15:50 15:50 2:30 other other other 16:40 16:40 16:40 bed 3:20 17:30 17:30 17:30 4:10 18:20 18:20 18:20 5:00 garage 19:10 garage 19:10 garage 19:10 bath 5:50 20:00 20:00 20:00 6:40 20:50 20:50 20:50 7:30 21:40 21:40 21:40 living 8:20 22:30 22:30 22:30 9:10 23:20 23:20 23:20 Figure B9. Average time spent in different locations within a home on weekends in 2019 for 10:00 10:50 178 dining 11:40 100% 90% 80% 100% 90% 80% 100% 90% 80% 70% 60% 50% 70% 60% 50% 70% 60% 50% 40% 30% 40% 30% 40% 30% Over 75 12:30 20% 10% 0% 20% 10% 0% 20% 10% 0% 13:20 office 0:00 0:00 0:00 14:10 0:50 0:50 0:50 15:00 1:40 1:40 1:40 15:50 other 2:30 2:30 2:30 16:40 bed 3:20 bed 3:20 bed 3:20 17:30 4:10 4:10 4:10 18:20 5:00 5:00 5:00 garage 19:10 bath 5:50 bath 5:50 bath 5:50 20:00 6:40 6:40 6:40 20:50 7:30 7:30 7:30 21:40 living 8:20 living 8:20 living 8:20 22:30 9:10 9:10 9:10 23:20 10:00 10:00 10:00 10:50 10:50 10:50 dining dining dining 11:40 65-74 11:40 45-54 11:40 25-34 12:30 12:30 12:30 13:20 13:20 13:20 office 14:10 office 14:10 office 14:10 15:00 15:00 15:00 15:50 15:50 15:50 other 16:40 other 16:40 other 16:40 17:30 17:30 17:30 18:20 18:20 18:20 garage 19:10 garage 19:10 garage 19:10 20:00 20:00 20:00 20:50 20:50 20:50 21:40 21:40 21:40 22:30 22:30 22:30 23:20 23:20 23:20 people of different age group 6. CHAPTER 6 – CONCLUSION AND FUTURE WORK 1. Conclusions In this study, typical individual occupancy profiles for U.S. residential buildings are created using ATUS and RECS data and compared with the schedules used in current energy modeling methods and tools. The schedules were then mapped to typical households of different sizes. The spatial distribution of occupants in indoor residential spaces was evaluated based on the time of day and the percentage of time people spent in different rooms. Next, cluster analysis was completed to determine patterns in occupancy schedules for people across different age groups and whether it is weekday or weekend. After that, the accuracy of the cluster profiles was evaluated and the variations in the profiles were studied. Different characteristics of the schedules were also analyzed. Next, three household income groups, including low-income, medium-income and high-income households, and 7 age groups, were analyzed. The average occupancy profiles were evaluated and then cluster analysis was implemented using these profiles. The typical age combinations of households by household size were also determined based on RECS data, resulting a comparison of occupancy profiles for the most common age combination by income level. The remaining portion of this research focuses on quantifying the impact of the pandemic on occupancy and activities occurring in homes using a combination of ATUS and CPS data for 2020, compared to the pre-pandemic years of 2006 to 2019. The impact of different time, occupant-related, and household characteristics on occupancy during the pandemic was evaluated and compared to pre- pandemic times. For time variables, the variation in weekday and weekend profiles and across months were evaluated. For household characteristics, the differential impact across different household incomes and household sizes were assessed. Finally, the activities and indoor locations used throughout a typical day were compared on weekdays and weekends, to pre-pandemic times. The overall key findings can be summarized as follows: • Across the years of data (pre-2020) analyzed, occupancy schedules do not change significantly. • Overall, approximately 75% of people’s (in the U.S.) time is spent in residential buildings. 179 • The typical individual occupancy schedules vary substantially based on the age of the occupants and whether it is a weekday or weekend. • The variations in typical individual occupancy schedules among different age groups are much higher on weekdays compared to that in weekends. For people over 65, occupancy profiles remain similar on both weekdays and weekends. • The average occupancy fraction of individuals in older age groups is higher compared to individuals in younger age groups. In addition, the average occupancy fraction of individuals on weekends is higher compared to weekdays. On weekdays, the minimum occupancy fraction for older age groups is approximately 0.3 higher compared to younger age groups. On weekends, the difference is smaller, approximately 0.1 among different age groups. • The occupancy schedule profiles used in the DOE Residential Reference Building and Building America (BA) Simulation Protocol overestimate the occupancy from 5:00 to 8:00 am. These underestimate occupancy from 7:00 to 10:00 pm compared to the typical individual occupancy schedule developed herein. • Overall, the trends of the typical household occupancy schedule profiles are most similar to those currently used in the DOE Reference Building and Building America (BA) Simulation Protocol for the 3- and 4-person households, however there are larger differences between the currently used schedules and the 1- and 2- members household schedules. • The distribution of amount of time that people are absent from their home on weekdays and weekends varies across age groups. Around 42 to 44% of occupants under 55 are absent from their home for 8 to 12 hours periods, whereas, for those 55 to 64, only 33% of occupants are absent for this period of time. For, people 65 and older, this value reduces significantly. • When at home, people spend a majority of their time in the bedroom (54-63% in weekdays and 55-62% in weekends) followed by the living room (19-27% in weekdays and 23-27% in weekends). • When considering the total time spent in different areas of a home, occupancy profiles are similar on weekdays and weekends. 180 • The average occupancy fraction of LIH individuals is significantly higher compared to MIH and HIH individuals. In some scenarios, specifically 10 am to 2 pm, the average occupancy fraction of HIH individuals on weekends is lower compared to that of weekdays in LIH individuals. Overall, midday, the average occupancy fraction of LIH is approximately 0.3 and 0.1 higher for weekdays and weekends, respectively, compared to HIHs. The occupancy fraction of MIH varies between these two profiles. • In total, 9 different types of occupancy profiles were obtained as a result of the cluster analysis, including (a) Stay home, (b) Day absence, (c) Day absence, less time, (d) Extended absence, (e) Day absence, earlier in day, (f) Day absence, later in day, (g) Absence very early in day, (h) Absence very late at night and (i) Night absence. Among these profiles, the Stay home profile was the most common for individuals on weekends. On weekdays, the Stay home profile is only the most common for LIH individuals, unlike HIHs, where Day absence or Extended absence profiles were most common. For MIH individuals, younger age groups generally followed one of the absence schedules, and the older age groups followed the Stay home schedule. • The occupancy fraction of households varies with respect to the number of household members and their overall household income. On weekdays, the occupancy fraction of the most common HIHs for 2-, 3- and 4-member households are 0.2 lower than the LIH. This difference is as high as 0.47 for 1-member households. The difference reduces on the weekend, from 0.07 to 0.18. • During the pandemic in 2020, people spent more time at home than pre-pandemic. Across this period, the average occupancy fraction on weekdays and weekends was 0.15 higher, and 0.1 higher throughout the majority of the day. Overall, per day, people spent around 1.9 and 1.2 hours more time at home on weekdays and weekends, respectively, which also varies with people’s age. On average, people also came home earlier on both weekdays and weekends during pandemic. • Considering variations across the different months of 2020, in the beginning of the pandemic (May) a maximum occupancy fraction difference of 0.3 was observed compared to pre-pandemic. This decreased slowly throughout the remainder of the year, to 0.15 in September, then remained approximately 0.2 for the last three months of 2020. 181 • For 2-, 3- and 4- member households, the pandemic resulted in an increase from 17.8 (pre- pandemic) to 19.9 hours of time spent at home on weekdays and 20 (pre-pandemic) to 21.1 hours at homes on weekends. 3- and 4-member households were more impacted. However, 1-person households were comparatively less impacted, where time spent in weekends were similar and on weekdays, they spent on average 32 minutes less in their home. On weekdays, the time at which people most commonly left their home was also delayed slightly. On weekends, people spent more time outside their home in the evening. • Across age groups, although different age groups follow different typical occupancy profiles, the overall impact was similar. On average, the occupancy dropped by 0.07-0.08 across a typical 24-hour day, with a maximum difference in occupancy fraction of 0.25 and 0.3 on weekdays and weekends respectively, typically occurring mid-day. • Across different household incomes, the average occupancy profiles on weekends were similar. However, on weekdays, people in low-income households spent less time at home during the day compared to the other two income groups. Pre-pandemic, people in higher income group spent the least time at home (~16 hours) and people from low-income households spent the maximum (~20 hours). This pattern reversed during the pandemic where people from low-income households spent around 18.6 hours at home which is 0.5 hours less than people from the high-income group. • Regarding activities and locations that people spent time in their homes, during the pandemic on weekdays, most time in the morning before noon was spent in office areas, primarily doing remote working and schooling; in the afternoon people continued to work and then transitioned more towards leisure activities, mostly corresponding to the living room area. On weekends, the living room and dining room/kitchen were the two most used spaces across all age groups, whereas younger age groups also used the office/study spaces during this time. Compared to pre-pandemic, the office and kitchen/dining space usage increased significantly whereas the living room and bedroom usage decreased. Thus, the analysis of the 15 years of time use survey data provides a detailed representation of occupant behavior and their schedules in residential buildings in the United States. This study showed that occupancy schedules pre-pandemic do not vary significantly across many years. Once the COVID-19 pandemic occurred in 2020, this forced people to change their lifestyles, resulting 182 in substantially different occupancy profiles as compared to the many prior years of data. In addition, along with the day of the week, age of the occupants also plays a significant role in shaping the usage profile of homes. Variation in occupancy schedule by household income is also notable, both pre- and during the pandemic. All developed profiles also are notably different from the currently used occupancy profiles in most energy modeling programs in use today. This suggests there is substantial opportunity to adjust occupancy schedules in energy modeling efforts to improve the accuracy of occupancy representation. This information can also be helpful to identify the demand response or energy reduction programs targeted to occupants with specific demographics, while having the least impact on occupant comfort. The indoor space usage profile also varies across different groups of occupants, which provides more insight for building designers and architects to design the spaces for the target household of specific demographics. The schedule and indoor location evaluation also provides the groundwork for designing a building occupancy simulator for residential households for occupants belong to different demographics. 2. Limitation There are a few limitations and assumptions associated with this study that are worthy of noting. One limitation is that the ATUS and RECS data have been used in combination. The reason the data was merged for use in this study is that ATUS data does not have information on the schedules of all household members, only the schedule for one person in the household (head of household). Therefore, an additional dataset is needed to define the other occupant(s)’ schedule(s) in the household and link multiple household members together. This is a limitation of the dataset that could be explored in future studies, such as through field-collected data. Another limitation of this study is that the utilized ATUS data is self-reported; self-reported data can include human error that may influence the results of this work. In addition, as the ATUS is self-reported data, it can be subjected to human error. Finally, the survey contains activity data for individual household members for individual, representative days. The information of activity data for multiple days and for multiple household members is not currently available using existing nationwide survey mechanisms. However, this would be helpful for future work to better understand variation in 183 occupancy schedules within households. In addition, some activities reported in the ATUS also include location information while others are assumed based on the nature of the activity. 3. Research contribution This study focuses on the characteristics of occupancy profiles in the residential building in the United States and analyze the impact of various parameters. This study analyzed the residential occupancy scenarios for the overall United States for a span of 15 years, a length of time that was not analyzed previously. These schedules are studied for different occupant characteristics such their age group how they change across weekdays and weekends. Additionally, in a novel approach, occupancy schedules are divided based on number of household members and the profiles are developed for each of the household types and sizes individually. This is important as the residential buildings generally have on/off type HVAC controls and thus the overall household profiles provide more insight about the potential savings that can be achieved in different types of households. This study is unique as there were no previous studies that use these demographic factors based on both occupant/household characteristics as well as building characteristics to evaluate the residential occupancy profiles. This study is a significant step forward in occupancy schedule estimation for residential buildings from what is currently in use for most of the building simulation platforms, reference building models, and standards. This study creates a different dimension of insights as to how residential occupancy profiles should be assumed to evaluate the energy consumption in residential buildings. Indoor space usage was also studied to provide more insights about how a typical residential building is used. Another important factor in this study is the variation in occupancy profiles by household income. The mapping of the number of household members and their household income and the impact on profile variations were not studied previously. This study shows that generally occupants that belong to lower income groups stay at home the most compared to others. This shows the importance of proper policy design for lower income households rather than a single demand response program configuration to reduce their energy burden. Another unique dimension that is studied in this dissertation is the impact of pandemic on the residential occupancy profiles. The impacts are evaluated across various occupant and household related variables and compared with the pre-pandemic time-period. This study provides a better 184 understanding of how substantial an impact the pandemic has had on house people spend their time, in particular in their homes. These results will help to better understand how buildings will be used as the pandemic continues. 4. Future work As a future study, using these parameters, an occupancy simulation tool can be generated which can generate stochastic annual occupancy schedules for different types of residential buildings. In addition, occupancy-dependent energy end uses, such as appliances and plug loads, can further benefit from, and be updated based on the finding of this research, and further analysis of ATUS and other related and complimentary datasets. With additional linked data on how such households use energy consuming devices, additional impacts of occupancy on energy consumption can also be studied. In addition, as the profiles become more consistent as the pandemic continues, and as newer ATUS data becomes available, it would be insightful to link the 2020 data to newer data to understand longer term trends and how they vary in the various stages of the pandemic. In addition, a better understanding of the spatial location distribution is useful for optimizing the deployment of occupancy sensor systems to detect occupant in a residential building. These can be used as input into residential building energy simulation scenarios to assess income and household size impacts on building energy performance, in particular for devices that control buildings systems such as lighting and HVAC systems based on whether or not the building is occupied. 185 7. CHAPTER 7 – PUBLICATIONS 1. Journal Publications 1. Mitra, D., Chu, Y., and Cetin, K. S., (2022) COVID-19 impacts on residential occupancy schedules and activities in US Homes in 2020 using ATUS. Applied Energy (2022): 119765. 2. Mitra, D., Chu, Y., Cetin, K.S., Wang, Y., Chen, C.F. (2021) Variation in Residential Occupancy Profiles in the United States by Household Income Level and Characteristics. Journal of Building Performance Simulation: 14(6), 692-711 3. Mitra, D., Chu Y., Cetin, K.S., (2021) Cluster analysis of occupancy schedules in residential buildings in the United States. Energy and Buildings 236: 110791 4. Mitra, D., Steinmetz, N., Chu, Y., and Cetin, K. (2020) Typical occupancy profiles and behaviors in residential buildings in the United States. Energy and Buildings 210: 109713 2. Book Chapter 1. Mitra, D., Malekpour, D., Cetin, K.S. Chapter 6: Occupancy Data Sensing, Collection, and Modeling for Residential Buildings. Renewable Energy for Buildings, Springer Nature (accepted) 3. Conference Proceedings 1. Mitra, D., Younessi, T., Kawka, E., Cetin, K.S., (2022). Variations in Residential Building Electricity- Consuming Appliances due to the COVID-19 Pandemic, ASHRAE Annual Conference, Toronto, on June 25-June 29, 2022 2. Mitra, D., Chu Y., Cetin, K.S. (2022), Impact of Occupant Characteristics on the Energy Performance of Multifamily Residential Building in the United States, 6th Residential Building Design & Construction Conference, State College, PA, March 2-3, 2022 3. Mitra, D., Chu Y., Cetin, K.S., (2021). Characteristics of Residential Occupancy Profiles for Different Income Groups in the United States, 2021 ASHRAE Winter Conference, Chicago, IL, January 23- January 27, 2021 4. Mitra, D., Chu Y., Cetin, K., (2020). Development of typical occupant profiles in academic buildings in the United States, 2020 ASHRAE Annual Conference, Austin, TX, June 27-July 1, 2020 5. Mitra, D., Steinmetz, N., Chu, Y., Cetin, K., (2020) Typical Activity Profiles of Occupants in Residential Buildings using The American Time Use Survey Data, Construction Research Congress 2020, Tempe, AZ, March 8-10, 2020 6. Mitra, D., Cetin, K., (2020) Optimum Properties and Orientation of Phase Change Materials for High Performance Buildings, Construction Research Congress 2020, Tempe, AZ, March 8-10, 2020 186 7. Mitra, D., Steinmetz, N., Chu, Y., Cetin, K., (2020) Characteristics of Typical Occupancy Schedules for Residential Buildings in The United States, 5th Residential Building Design & Construction Conference, State College, PA, March 4-6, 2020 8. Mitra, D., Chu, Y., Steinmetz, N., Cetin, K., Lovejoy, J., Kremer, P. (2019) Defining Typical Occupancy Schedules and Behaviors in Residential Buildings Using the American Time Use Survey, 2019 ASHRAE Annual Conference, Kansas City, MO, June 22-26, 2019 4. Others 1. Mitra, D., Cetin, K., Edil, T. B., Cetin, B., (2021), Environmental Impacts on the Performance of Pavement Foundation Layers, Department of Transportation, Minnesota. 2021/8/1 2. Chu, Y., Mitra, D., Cetin, K., Lajnef, N., Altay, F., Velipasalar, S., (2022) Development and testing of a performance evaluation methodology to assess the reliability of occupancy sensor systems in residential buildings. Energy and Buildings 268: 112148. 3. Chu, Y., Mitra, D., O’neill, Z., & Cetin, K. (2022). Influential variables impacting the reliability of building occupancy sensor systems: A systematic review and expert survey. Science and Technology for the Built Environment, 28(2), 200-220. 4. Chu, Y., Mitra, D., & Cetin, K. (2020). Data-Driven Energy Prediction in Residential Buildings using LSTM and 1-D CNN. ASHRAE Transactions, 126(2), 80-88. 5. Kunwar, N., Vanage, S., Peruski, E., Mitra, D., & Cetin, K. (2020). Occupant-Dependent Residential End Use Load Profiles for Demand Response Under High Renewable Energy Scenarios. ASHRAE Transactions, 126(2), 3-6. 6. Chu, Y., Evans, S., Mitra, D., & Cetin, K. (2022) Experimental Design and Analysis of Reliability Evaluation of Off-the-Shelf Occupancy Sensor System in Residential Buildings, ASHRAE Transactions. 2022, Vol. 128 Issue 1, p331-339. 9p. 7. Chu, Y., Mitra, D., & Cetin, K. (2021). Implementation of Occupant-based Control in Typical Academic Buildings, ASHRAE Transactions. 2021, Vol. 127 Issue 1, p152-160. 9p. 187