Personalized cooling as an energy efficiency technology for city energy footprint reduction

Journal of Cleaner Production 171 (2018) 491e505

Contents lists available at ScienceDirect

Journal of Cleaner Production journal homepage: www.elsevier.com/locate/jclepro

Personalized cooling as an energy efficiency technology for city energy footprint reduction Mohammad Heidarinejad, Daniel Alejandro Dalgo, Nicholas W. Mattise, Jelena Srebric* Department of Mechanical Engineering, University of Maryland, College Park, MD, 20742, USA

a r t i c l e i n f o

a b s t r a c t

Article history: Received 15 May 2017 Received in revised form 24 September 2017 Accepted 2 October 2017 Available online 3 October 2017

This study analyses the influence of Personalized Conditioning (PC) systems for potential savings of energy, cost, and CO2 emissions from commercial buildings in different U.S. cities. This analysis characterizes potential benefits from the deployment of PC systems during peak cooling hours for peak load shifting. PC systems deployed in coordination with the central building air conditioning systems could have a large-scale influence on a city's energy footprint. Specifically, portable PC systems that use Phase Change Materials (PCMs) for heat rejection, allow for heat absorption during the working hours and heat rejection during non-working hours typically coinciding with the off-peak (base) utility rates when the commercial building tend to be unoccupied. However, there are limiting factors for the potential energy and cost savings with the use of PC systems. Therefore, this study assesses the use of PC systems in addition to the existing building air conditioning system during cooling seasons. The assessment entails potential energy end-use savings for 7 major cities located in different geographical/climatic regions of the U.S. Furthermore, the study calculates potential cost savings based on the variations in the peak and off-peak (base) electricity rates for different local Time of Use (TOU) programs. This simulated evaluation of local building systems and utility programs allows for regional various on the city's energy footprint reduction to emerge. The analysis shows that midrise apartments are a better building type than office buildings for the deployment of PC systems during a cooling season. The cash savings per person for the deployment of PC systems for midrise apartments are $62/year, $40/year, and $37/year for Honolulu, NY City, and Phoenix, respectively. The simulations also showed that using extended setpoint temperatures could reduce the CO2 emissions up to 21.4% per year. © 2017 Elsevier Ltd. All rights reserved.

Keywords: Personalized conditioning Building energy savings Thermal comfort Utility rate

1. Introduction Improvement of energy efficiency in the building sector is the most effective strategy to mitigate the impact of people on the built environment and to match the energy and power production with the demand. Residential and commercial buildings account for 41% of source energy in the U.S. (Commercial Building Energy Consumption Survey (CBECS), 2012). Commercial and residential buildings account for 9.9% and 5.4% global Greenhouse Gas Emission (GHG) emissions, respectively (Shen, 2017). The U.S. is responsible for 21% of the world's CO2 emissions, and 98% of the U.S. emissions are from the energy consumption (Attari et al., 2010). Recent studies elaborated the serious side effects of CO2 emissions on national economies due to global warming and climate change

* Corresponding author. E-mail address: [email protected] (J. Srebric). https://doi.org/10.1016/j.jclepro.2017.10.008 0959-6526/© 2017 Elsevier Ltd. All rights reserved.

(Chang, 2015). It is expected that if the warming trend continues for the next 50 years it could increase the mean air temperature up to 1.3
C (Davis et al., 2010). This increase and other effects of global warming do not vary equally throughout the planet in time nor space, suggesting regional extreme temperature conditions could follow different patterns than the global warming (Shepherd, 2015). For example, a study has shown that the China's CO2 peak emission could occur prior to 2030 with careful considerations of GHG emission mitigation strategies based on limits to Gross Domestic Product (GDP) and energy/CO2 intensity (Mi et al., 2017). Importantly, the GHG emission rates could be resolved to different spatial scales varying from the city to the globe (Meng et al., Liu). Consequently, the influence of the energy systems on climate change requires regional strategies. Retrofitting existing buildings to use significantly less energy is an essential part of any climate change mitigation and energy security strategy. Implementation of energy efficiency measures in existing buildings or design of energy efficient buildings are common approaches to reduce energy

492

M. Heidarinejad et al. / Journal of Cleaner Production 171 (2018) 491e505

Acronyms AHS BAS CAV CBECS CDD DOE FTE GHG HVAC LMP PC PCM RECS TOU VAV

American Housing Survey Building Automation System Constant Air Volume Commercial Building Energy Consumption Survey Cooling Degree Days The U.S. Department of Energy Full Time Employee Greenhouse Gas Emission Heating, Ventilation, and Air Conditioning Locational Marginal Pricing Personalized Conditioning Phase Change Material Residential Energy Consumption Survey Time of Use Variable Air Volume

consumption in the buildings or curb GHG emissions from the buildings (Dahlhausen et al., 2015). Although these approaches could support energy efficiency improvements, there is a need to move the boundaries beyond the current design and retrofit practices. One practical approach would be to benefit from the advances in the other disciplines and develop new technologies, especially for the space heating and cooling, to improve their energy efficiency (Raman et al., 2014). However, any disruptive technology requires original evaluation for economic survival strategy. This study considers advances in the design of energy efficient Personalized Conditioning (PC) systems to reduce the impacts of buildings on GHG emissions. Among the energy end-uses, space cooling and heating take up to 31% and 43% of the total building primary energy consumption (Huang and Gurney, 2016). Therefore, an effective deployment of PC systems considers the reduction of space conditioning in the building sector. These systems offer several benefits, including electricity peak load shifting, thermal comfort, and space air conditioning. Depending on the purpose of the PC systems, there are various PC heating and cooling systems available, including cooling and heated chairs (Brager et al., 2015), hand/foot warmers (Zhang et al., 2015), and portable Phase Change Material (PCM) supported systems (Dhumane et al., 2016a). Hand warmers, hand ventilation, foot warmers, head ventilation, and portable fans are among the most recent PC systems (Brager et al., 2015). The primary focus of PC systems is not on the reduction of energy consumption. Thermal comfort and increases in productivity are typically the primary factors in the deployment of these PC systems (Zhu et al., 2017). Consequently, thermal comfort satisfaction is not necessarily correlated with the associated energy consumption. Providing thermal comfort satisfaction of all occupants is not possible due to the individual differences, including age, gender, clothing, activity, and body mass (Van Hoof, 2008). It is even more difficult to satisfy all occupants using the centralized systems. Consequently, one practical approach is to create a local thermal comfort zone for the building's occupants and address individual complaints using PC heating and cooling systems. Among the energy consumption commodities in buildings, space cooling is a significant portion of building energy consumption patterns. According to Commercial Building Energy Consumption Survey (CBECS) 2012, space cooling in the U.S. requires 19% of the total electricity consumption (Commercial Building Energy Consumption Survey (CBECS), 2012). In addition,

among the commercial building types, offices account for 17% and 19% of the commercial building floor area and primary energy consumption, respectively (2010 Buildings Energy Data Book, 2011). Therefore, among commercial buildings, office buildings are a good starting point for energy saving technologies (Heidarinejad et al., 2014). For residential buildings, Residential Energy Consumption Survey (RECS) and the American Housing Survey (AHS) are sources of data with the focus on data from 1970 to 2011 (Moura et al., 2015; Belzer, 2014). The total floor area for residential buildings increased from 87 million square feet to 211 million square feet from 1970 to 2011 (Belzer, 2014). Therefore, one promising strategy to reduce the energy consumption in buildings is to consider reduction in space cooling through no-cost/low-cost building operation energy efficiency measures to deployment of novel technologies. Increasing the cooling setpoint, decreasing the heating setpoint, and decreasing the ventilation rate of the building's primary Heating, Ventilation, and Air Conditioning (HVAC) systems are among the main assumptions to save energy consumption (Heidarinejad, 2014). The suggested change for the setpoint is to decrease the heating setpoint or increase the cooling setpoint of the primary HVAC system to 4e5 K (Hoyt et al., 2015). The current state-of-theart showed up to 60% energy savings (Veselý and Zeiler, 2014). These assessments are mainly based on the use of building energy modeling that is a powerful tool to assess potential energy and cost savings as well as GHG emission mitigation associated with implementing different saving strategies. Consequently, this study benefits from building energy modeling to evaluate potential energy and cost savings as well as GHG emission reduction at a largescale. The validation and verification of the building energy models requires relying on a systematic calibration of the building energy models. The authors have demonstrated using building energy modeling for understanding office building occupancy patterns (Kim et al., 2017), building energy retrofits of office buildings (Dahlhausen et al., 2015), and quantification of the urban microclimate on a multi-family residential building (Liu et al., 2015). Significant impacts on the energy efficiency of buildings require a large-scale analysis at the neighborhood, city, or nation scale. Cities are living laboratories since they are located in different geographic locations and have different policies. However, the deployment of PC systems at the city scale is expensive and mostly impractical with the current resources. This study suggests a much less expensive approach to perform the initial analysis computationally with validated numerical models and provide suggestions for the large-scale deployment of these PC systems in various cities in the U.S. City scale modeling requires consideration of energy market constrains, e.g. demand response programs, Locational Marginal Pricing (LMP), and Time of Use (TOU). The aim of these programs for the utility providers is to provide (1) electricity load shifting from the peak hours to off-peak hours (Qiu et al., 2016) and (2) opportunities for the customers to reduce their electricity bill cost (Kirkeide, 2012). These PC systems could offer additional features, e.g. heatwave conditions when the peak load on the power grid increases drastically (Hannah, 2015). Consequently, this study considers seven different cities that have different local climates and energy markets so there are opportunities to test the implementation of different mitigation technologies with the support of PC systems. The current study considers energy and cost saving features, including laying out assumptions based on the current-state-ofthe-art for PCM supported portable PC systems that could offer several benefits from potential energy and cost savings to thermal comfort. Overall, the aim of this study is to: 1. Quantify the potential energy end-use and cost savings for office and midrise apartment buildings located in different climate

M. Heidarinejad et al. / Journal of Cleaner Production 171 (2018) 491e505

zones in the U.S. using extended setpoints and PC systems during cooling months. 2. Investigate the influence of TOU programs for the utility rates for 7 U.S. cities to demonstrate a wide range of applications. 3. Provide estimation for the potential energy savings for the residential building types for the entire U.S. building stock. 4. Summarize the influence of the PC systems for a large-scale city deployment to reduce city energy footprints and potential climate change mitigation strategies.

493

These spaces represent building types that include Full Time Employees (FTEs) or residents, rather than a large population of transient occupants such as the ones in retail buildings. The selection of these two building types empowers this study to assess the relative significance of building consumption patterns in terms of their response to the outdoor conditions. While office buildings tend to be internally-load dominated, meaning the internal loads are the major driver for an office building energy consumption patterns, midrise apartments are externally-load dominated buildings, meaning that the primary heat transfer in the buildings depends on the outdoor conditions (Heidarinejad et al., 2017). Overall, the selection of these two building types allows quantification of the relative significance of the heat transfer processes in buildings.

2. Research methodology This section provides the research methodology used in the study. One approach to benefit from the PC systems is to use PCM, allowing the absorption of heat during the day from the space and rejecting heat during the night time. There are two methods to reject the heat during the night time: (1) reject heat to the internal spaces during the night when the building is unoccupied and the building load is not significant and (2) reject heat outside of the building. Since the regeneration temperature for the PCM during cooling season is high enough, it is possible to benefit from the TOU programs for the utility rate and indoor temperature to regenerate the PCM of the PC systems after the working hours. The steps in this section are to: (1) Select different geographic locations, (2) Consider the most common and significant building type contributors to the building sector energy consumption, and (3) Detail modeling approach, (4) Select different utility rates and utility purchasing options on the application of PC systems, and (5) GHG calculation assumptions.

2.3. Building energy models This study uses the U.S. Department of Energy (DOE) Reference Buildings (Deru et al., 2011) as the baseline for the building energy models. The selection of office and midrise apartment building principal activities allows an assessment of the influence of the building occupancy patterns, temperature setpoints, control strategies, and HVAC systems. These two types of building represent building types that have FTEs or residents. Therefore, when they are in the building, they spent a significant amount of time in the space. In addition, the building construction, infiltration, ventilation rate requirements, and building HVAC age are varied as this study considers two different buildings represented with (1) Pre-1980 information and (2) New Construction specifications. The office and midrise apartment buildings have 268 and 58 FTEs and residents, respectively. The area of these the midsized office and midrise apartments prototype buildings are 4,892 m2 and 2,030 m2, respectively.

2.1. Geographic locations Major factors in the selection of the representative cities in this study are: (i) Assess impacts of different climate conditions, (ii) Consider different utility cost programs, and (3) Enable future discussions on utility rebate programs. Table 1 lists the selected cities and associated climate zones based on the ASHRAE Standard 169e2006 Standard (ASHRAE Standard 169-2006, 2006). The cities cover different climate conditions and the focus is on the humid locations denoted with “A” and “C” that are more demanding in terms of PC system deployment. The selection of these representative major cities allows considering different climate condition in order to estimate the energy savings for the city's climate condition. While the cities of Austin, Phoenix, and Honolulu are cooling dominated, the other four cities require minimal or less cooling energy compared to the cities located in warm climates. Each city has dominant utility providers that could serve as a source for the utility rates. Another reason for this selection was to consider variation in the TOU programs and associated utility costs. Overall, this selection allows the considering of various strategies for the potential reduction in the city energy footprints.

2.4. TOU utility rates The utility rate variation from the peak hours to the off-peak hours could have a significant effect on the electricity cost assessment. Since the difference between the peak and base electricity price could be up to twenty times, suggesting opportunities to benefit from the peak load shifting strategies. Existing studies in the literature indicate that TOU programs lead to significant reduction in the peak hour electricity consumption while the overall electrical consumption remain unchanged under these programs (Jessoe and Rapson, 2015). Consequently, one of the best applications for the deployment of PC systems is for the utility providers to suggest rebate programs in order to reduce the annual peak load and provide cost savings for their customers. With the recent advances of using PCM supported portable heating and cooling systems, it is possible to charge the devices during the off-peak hours and deploy the systems during the peak hours to facilitate the peak load shifting strategies. These devices have internal thermal inertia to absorb heat during the day without any additional energy use and be regenerated during the off-peak hours when the electricity price is lower. Therefore, it is vital to consider the variable utility cost for energy commodities during the day and night for the selected representative cities.

2.2. Building types This study proposes to start deployment of the PC systems in two main building types: (a) offices and (b) midrise apartments.

Table 1 The selected climate zones and representative cities in this study. City Number

1

2

3

4

5

6

7

Representative City Climate Zone

Austin, TX 2A

Chicago, IL 5A

Honolulu, HI 1C

Minneapolis, MN 6A

New York City, NY 4A

Phoenix, AZ 4B

San Francisco, CA 3C

494

M. Heidarinejad et al. / Journal of Cleaner Production 171 (2018) 491e505

Table 2 Summer TOU utility costs and hours for commercial buildings in 7 major cities (cents/kWh). City \ Rate

On-Peak

Mid-Peak

Off-Peak (Base)

Months

Austin (City of Austin electric tariff, 2016)

6.54 (2 p.m.e8 p.m.)

0.67 (10 p.m.e6 a.m.)

Jun. to Sept.

Minnesota (Minnapolis commercial utility rates, 2016) Honolulu (Hawaii commerical utility rates, 2016) New York City (New York City commercial utility rates, 2016) Chicago (New York City commercial utility rates, 2016) San Francisco (San Francisco commercial utility rates, 2016)

15.13 (9 16.9 18.99 (8 18.99 (8 25.8 (12

15.41 (2 p.m.e7 p.m.)

3.02 (9 p.m.e9 a.m.) 16 1.34 (12 a.m.e8 a.m.) 1.34 (12 a.m.e8 a.m.) 20.7 (12 a.m.e8:30 a.m. &9:30 p.m.e12 a.m.) 5.48 (12 a.m.e11 a.m.)

Jun. to Sept. All Year Jun. to Sept. Jun. to Sept. May to Oct.

Phoenix (Summer) (Phoenix commercial utility rates, 2016)

May to Oct.

Phoenix (Summer Peak) (Phoenix commercial utility rates, 2016)

16.48 (2 p.m.e7 p.m.)

3.91 (6 AM e 2 p.m. & 8 PM e 10 p.m.) N/A 16.9 N/A N/A 23.4 (8:30 a.m.e12 p.m. &6 a.m.e9:30 p.m.) 10.50 (11 AMe2 p.m. & 7 PM e 11 p.m.) 10.70 (11 AMe2 p.m. & 7 PM e 11 p.m.)

5.15 (12 a.m.e11 a.m.)

July to Aug.

a.m.e9 p.m.) a.m.e12 a.m.) a.m.e12 a.m.) p.m.e6 p.m.)

Table 3 Summer TOU utility costs and hours for residential buildings in 7 major cities (cents/kWh). City \ Rate

On-Peak

Mid-Peak

Off-Peak (Base)

Months

Austin (City of Austin residential utility rates, 2016)

11.00 (2 p.m.e8 p.m.)

1.19 (10 p.m.e6 p.m.)

Jun. to Sept.

Minnesota (Minnapolis residential utility rates, 2016) Honolulu (Hawaii residential utility rates, 2016) New York City (New York City residential utility rates, 2016) Chicago (New York City residential utility rates, 2016) San Francisco (San Francisco residential utility rates, 2016)

20.00 (9 a.m.e9 p.m.) 39.2 (5 a.m.e10 p.m.) 20.53 (8 a.m.e12 a.m.) 20.53 (8 a.m.e12 a.m.) 30.0 (3 p.m.e8 p.m.)

6.22 (6 a.m.e2 p.m. & 8 p.m. 10 p.m.) N/A 25.6 (10 p.m.e9 a.m.) N/A N/A N/A

Jun. to Sept. All Year Jun. to Sept. Jun. to Sept. Jun. to Sept.

Phoenix (Phoenix residential utility rates, 2016)

24.47 (12 p.m.e7 p.m.)

N/A

3.02 (9 p.m.e9 a.m.) 16.8 (9 a.m.e5 p.m.) 1.45 (12 a.m.e8 a.m.) 1.45 (12 a.m.e8 a.m.) 22.0 (12 a.m.e3 p.m. & 8 p.m.e12 a.m.) 6.11 (7 p.m.e12 p.m.)

Tables 2 and 3 show the TOU utility rates and associated hours for the peak, mid-peak, and off-peak for summer time for seven major cities in the U.S. This study assumes similar utility rates for both cities of New York and Chicago since they have similar utility provider. The commercial and residential TOU rates suggest that in terms of the variation in the rates, there are two major patterns: (1) Cities, e.g. Honolulu and San Francisco, with fairly flat electricity pricing during the day regardless of the peak or offpeak hours. (2) Cities, e.g. Austin, New York City, and Chicago, with significant variation in electricity pricing from the off-peak (base) to peak hours. In addition, a comparison between the peak hour and off-peak hours indicate that there are cities with 12 h durations for the peak hours while there are cities with only 4 h for the peak hours. These variations influence the control strategies to benefit the PC systems since the deployment and regeneration hours are different. 3. Assumption of the analytical models This section provides details about the battery powered portable PC cooling systems, assumptions used in the building energy models, and the utility cost for different geographic locations. 3.1. Operation scenarios For each building type and city, there are five different setups: (1) baseline, (2) hard-sized baseline, (3) extended setpoint, (4) PC systems are placed inside for regeneration, and (5) PC systems are placed outside for regeneration. The baseline model is the DOE Reference building model. This study uses the baseline model to identify the size of equipment and then fix the size of equipment in the hard-sized baseline model. The extended setpoint is potentially the maximum achievable cooling for the defined setpoint value.

May to Oct.

The extended setpoint model does not provide any cooling when the room temperature is less than the cooling setpoint. Hence, there might be occupants that do not feel thermally comfortable since the extended setpoint model assumes a setpoint temperature of 26.7
C. To address those dissatisfied occupants, this study suggests deployment of PC systems. While these PC systems absorb heat and remove heat from the space, they require regeneration of their PCM. There are two options for the regeneration: (i) place the PCM indoor space and (ii) place the PCM outdoor space. All of these two options have lower overall energy and cost savings compared to the extended setpoint model. This is the cost of making the system distributed vs centralized systems. Overall, the energy, cost, and CO2 results are included for the following scenarios:  Scenario 1 denoted with “Max Limit” that shows the results for the extended setpoint temperature.  Scenario 2 denoted with “PCs In” that considers the deployment of PCs and enable regeneration of the PCM inside of the space.  Scenario 3 denoted with “PCs Out” that considers the deployment of PCs and enable regeneration of the PCM outside of the space.

3.2. Loads and hours of operation Based on the thermal comfort analyses, a portable PCM supported PC device is able to absorb 165 Watts during the operation and rejects 210 Watts during the regeneration mode (Dhumane et al., 2016b). Additional requirements for charging the battery and fan operation are 70 Watts and 10 Watts, respectively (Du, 2016). Table 4 illustrates the operational hours of the portable PC devices in the office building, midrise apartment, battery charging, and PCM regeneration. Due to the losses and different time scales in the regeneration and recharging time, the heat balance for the duration of recharging, regeneration, and PC usage are different.

M. Heidarinejad et al. / Journal of Cleaner Production 171 (2018) 491e505

495

Table 4 Hours of operation for the PC systems, regeneration of PCM, and battery recharging.

3.3. Temperature setpoints Fig. 1 illustrates the temperature setpoints for the office and midrise apartment buildings during baseline, extended setpoint, and PC usage time. Fig. 1(a) and (d) illustrate the baseline setpoint temperature for the office and midrise apartment buildings, respectively. The office building baseline setpoints indicate that during working hours the setpoint temperature is 2.7
C different while for the midrise apartment the setpoint value is fixed throughout a day. Fig. 1(b) and (e) show the extended setpoint temperature that considers fixed extended setpoint throughout a day. Fig. 1(c) and (f) depict, the 6 h extended hours during the day for both office and midrise apartments when the PCs are in use. 3.4. Number of occupants Office buildings and midrise apartments have 268 and 58 people, respectively. The number of occupants dictates the number of required PC systems. Typically, one occupant requires one PC system. In addition, all of the occupants are not always dissatisfied

with their thermal environment. This study assumes 50% of the occupants require using PC systems, suggesting an office building model and midrise apartment model have 134 and 29 PC systems, respectively. This is due to the fact that not all the occupants are dissatisfied with the indoor thermal environment. 4. Energy assessment in the major cities Fig. 2 illustrates the energy assessment for the selected cities for Scenario 1 and Scenario 3. Scenario 1 (Max Limit) is when the temperature setpoint is extended and there are no PC systems in the building. Scenario 3 is for the case when the PC systems are placed outside during the regeneration of the PCM. While the left yaxis in the figure shows the percent savings for fan, cooling, or heating energy, the right y-axis represents the total energy savings. Cooling energy represents 35% of a building's total for Austin, Honolulu, and Phoenix and 20% for Chicago, NY City, and Minneapolis. While the fan energy consumption accounts for only 3% of the total energy for new construction buildings, this amount is 17% for the old office buildings. Due to the inherent HVAC system

Fig. 1. Temperature setpoints for different baseline, extended setpoint, and PC operation: (a) office building baseline, (b) office extended setpoint, (c) office PC setpoint, (d) midrise apartment baseline, (e) midrise apartment extended setpoint, and (f) midrise apartment PC setpoint.

496

M. Heidarinejad et al. / Journal of Cleaner Production 171 (2018) 491e505

Fig. 2. Cooling, internal equipment, fan, and total energy savings for (a) Old office buildings for extended setpoint scenario, (b) New office buildings for extended setpoint scenario, (c) Old office building for PCs scenario, and (d) New office building for PCs scenario.

differences, there could have significant influence on the energy saving assessment of the PC systems. Old office buildings used Constant Air Volume (CAV), and the new construction office buildings use Variable Air Volume (VAV) systems. Internal equipment accounts for 27% and 35% of Pre-1980 and new construction buildings, respectively. As Fig. 2(a) indicates that percent while there are significant cooling energy savings, there are negative fan savings. Therefore, because 17% of the total energy is associated with the fan energy, the total energy savings for the old office building during the Max Limit Scenario is about 21% for cities such as Phoenix. Fig. 2(b) shows that the significance of total energy savings is lower for the new office buildings than the old office buildings. Phoenix and Honolulu show the highest amount of total energy savings. Fig. 2(c) and (d) depict the energy breakdown savings for the cooling, fan, internal equipment, and total energy. While for the old office buildings, the cooling savings are more significant than the new office buildings, the increase in internal equipment due to the use of PCs and fan energy lead to lower energy savings. The energy expense of distributed cooling requires consideration of the expense for the cooling, fans, and internal equipment. Fig. 3(a) and (b) depict the total energy savings when the PC's heat rejection takes place inside or outside for the regeneration of PCM during off-peak hours. The comparison shows that the results for both old and new office buildings follow the same pattern in terms of total energy savings while the magnitudes are different. A physical representation of the outside heat rejection for PCs would require ducted or un-ducted connections for PC's outside air intake and exhaust during the PCM regeneration. With the placement of PC's heat rejection outside, the total energy savings increased to 2%. All the cities show increased total energy savings for the PC's heat

rejection outside compared to the savings for the PC's heat rejection inside. Therefore, for the outside heat rejection from the PCM, there is no need to account for the rejected heat into the building during the night. Fig. 4(a) and (b) provide the comparison between the amount of cooling and fan energy savings. Among the cities, Phoenix shows the most savings when PCs are placed both inside and outside. For the old office buildings, due to the use of CAV systems, fans consume up to 18% more energy with the use of PCs than the baseline model. Consequently, for the old office buildings, the significant cooling savings with the use of PCs is drastically being offset by the higher fan energy consumption. Fig. 4(a) provides the distributions when the PCs are placed inside and when the PCs are placed outside. However, the new office buildings, depicted in Fig. 4(b), benefited from VAV systems that enable fan savings with the use of PCs. Although there are fan savings, the amount of cooling and fan energy savings are lower than the old office building overall cooling savings. Therefore, the expense of distributed cooling for the old office buildings require savings in the fan energy that could significantly increase potential energy savings from the deployment of PCs. Fig. 5 depicts the energy use patterns for the midrise apartments. Both Pre-1980 and new construction buildings use split CAV HVAC systems. Similar to Fig. 2, the left y-axis of Fig. 5 represents the fan, cooling, or heating energy savings, and the right y-axis of Fig. 5 illustrates the total energy savings. For the Pre-1980 buildings, cooling energy accounts for 40% of the total energy consumption for the buildings located in Austin, Honolulu, and Phoenix while cooling energy takes 17% of the total energy consumption for buildings in Chicago, NY City, and Minneapolis. Fan and internal equipment account for 7% and 25% of the total energy

M. Heidarinejad et al. / Journal of Cleaner Production 171 (2018) 491e505

497

Fig. 3. Total energy savings when the PC's heat rejection is inside (PCs In) and when the PC's heat rejection is outside (PCs Out) for: (a) Old office buildings, and (b) New office buildings.

Fig. 4. Cooling and fan energy savings when the PC's heat rejection is inside (PCs In) and when the PC's heat rejection is outside (PCs Out) for: (a) Old office buildings, and (b) New office buildings.

498

M. Heidarinejad et al. / Journal of Cleaner Production 171 (2018) 491e505

Fig. 5. Cooling, internal equipment, fan, and total energy savings for (a) Old midrise apartment buildings for extended setpoint scenario, (b) New midrise apartment buildings for extended setpoint scenario, (c) Old midrise apartment building for PCs scenario, and (d) New midrise apartment building for PCs scenario.

consumption in these midrise apartment buildings. Fig. 5(a) and (b) show the total energy savings for old and new midrise apartments, respectively. Both building types provide a high energy savings in all end-uses specifically the cooling energy with the consideration of extended setpoint. However, Fig. 5(c) and (d) indicate a placement of PC systems inside or outside could lead to lower energy savings comparted to the extended limit scenario. Overall, due to the higher contribution of internal loads with the use of the PC systems, the significant cooling and fan savings offset the total energy savings. Austin and Phoenix are among the cities that show a 9% and 7% total energy savings. Fig. 6 illustrates the total, cooling, and fan energy savings with the deployment of PCs and regeneration of the PCMs with the air from inside or outside of the building. Fig. 6(a) and (b) indicate the total energy savings vary from 0% to 9% for the old midrise apartment buildings and from 0% to 7% for the new midrise apartment buildings, respectively. The PC's heat rejection outside leads to 1%e 2% of additional energy savings compared to the savings with the PC's heat rejection inside, which is a similar finding to the one for office buildings. Furthermore, while the new office buildings for most the cities show a negative saving for the PC's heat rejection inside, the midrise apartments in all the cities and configurations resulted in additional energy savings. Importantly, the savings for the midrise apartments are approximately twice the savings for the office buildings. The calculated total energy savings from the use of PC systems in the midrise apartments compared to the savings in office buildings indicates that the midrise apartments are better candidates than the office buildings for the use of this specific PC system. Similar to Fig. 4 that illustrates the distribution of the cooling and fan energy savings for the placement of the PC's heat rejection

outside or inside, Fig. 7(a) and (b) depict the overall patterns for cooling and fan energy savings for both the placement of PC's heat rejection inside and outside. The savings for both placements are similar, and they vary from 12% to 47%. These results also explain why in Fig. 6, the new midrise apartments have positive net total energy savings. This is because there are significant fan energy savings in addition to the expected cooling energy savings. 5. Cost assessment in the major cities Use of TOU allows for benefiting from the difference between utility rates during peak hours and off-peak hours to deploy new scenarios. Fig. 8 summarizes the electricity cost savings for the PC systems for office buildings during the cooling seasons. Extended setpoint represented with “Max Limit” for most of the cities could lead to savings of thousands of dollars. In general, as Fig. 8(a) and (b) illustrate deployment of PC systems in old office buildings provides higher electricity cost savings compared to the new office buildings. Fig. 9 shows the electricity cost savings in midrise apartments. The Max Limit scenario shows up to $7,560 and $4,412 during summer season for old and new midrise apartments. Fig. 9(a) indicates the extended setpoint temperature has the most influence for Honolulu due to the fact that this increase in the setpoints lead to minimal cooling supply during the summer season. Among the cities, Honolulu, Austin, Phoenix, and NY City show a better savings compared to the other cities. For the new midrise apartments, Fig. 9(b) shows the savings for the new midrise apartments are lower compared to the old midrise apartment buildings. Still Honolulu, Phoenix, NY City, and Austin are among the most promising cities for the deployment of PC systems.

M. Heidarinejad et al. / Journal of Cleaner Production 171 (2018) 491e505

499

Fig. 6. Total energy savings when the PC's heat rejection is inside (PCs In) and when the PC's heat rejection is outside (PCs Out) for: (a) Old midrise apartment buildings, and (b) New midrise apartment buildings.

Fig. 7. Cooling and fan energy savings when PC's heat rejection is inside (PCs In) and when PC's heat rejection is outside (PCs Out) for: (a) Old midrise apartment buildings, and (b) New midrise apartment buildings.

500

M. Heidarinejad et al. / Journal of Cleaner Production 171 (2018) 491e505

Fig. 8. Electricity cost savings for the selected cities: (a) Old office, and (b) New office.

Tables 5 and 6 provide the annual cash savings in dollars per person for the old and new office and midrise apartments for the three options. Old offices tend to have higher cash savings per person compared to the new offices. Table 6 provides the annual cash savings in dollars per person for old and new midrise apartments for the three options. The extended setpoint scenario shows the highest potential cash savings since it does not add any internal loads to the building. With the introduction of the PC systems, the potential savings reduce due to the increase in the internal equipment energy use. All of the reviewed cities indicated a cash savings from $1 to $130 for annual cooling $/person, suggesting a potential for the widespread deployment of these PC systems. Among the energy end-uses, the cooling energy use is the highest contributor to the total energy use. Consequently, the derived cash savings for cities located in moderate or cold climates, e.g. Chicago and San

Francisco, are not as significant as the other cities due to the low cooling requirement. For most cities, the utility prices for the residential are higher than the commercial buildings. The midrise apartment showed a higher energy and cost savings compared to the commercial buildings. Overall, internal equipment and heat rejection during night time are the limiting factors for the amount of cash savings. In order to extend the applicability of the results, there is a need to develop quantitative metrics. Therefore, this study defined a price ratio as an indicator of the base to peak electricity price spread:

Price Ratio ¼ 1 

Base Price Peak Price

(1)

Combination of the price ratio and the degree days could serve

Fig. 9. Electricity cost savings for the selected cities: (a) Old midrise apartment, and (b) New midrise apartment.

M. Heidarinejad et al. / Journal of Cleaner Production 171 (2018) 491e505

501

Table 5 Annual cash savings for cooling ($)/person for the old and new offices.

Max Limit PCs Out PCs In

Old Office New Office Old Office New Office Old Office New Office

Austin

Chicago

Honolulu

Minneapolis

NY City

Phoenix

San Francisco

6 4 5 3 5 3

4 2 1 0 0 1

10 8 3 1 3 1

9 4 7 2 7 2

11 7 8 3 5 2

12 12 12 10 11 9

6 3 0 4 1 3

Table 6 Annual cash savings for cooling ($)/person for the old and new midrise apartments.

Max Limit PCs Out PCs In

Old Midrise Apt New Midrise Apt Old Midrise Apt New Midrise Apt Old Midrise Apt New Midrise Apt

Austin

Chicago

Honolulu

Minneapolis

NY City

Phoenix

San Francisco

60 35 32 22 30 21

19 12 6 4 6 3

130 76 62 38 58 35

42 25 22 16 21 15

78 78 40 40 36 36

64 38 37 27 36 26

29 23 3 4 1 2

Fig. 10. Price ratio and degree days for the selected seven cities.

as quantitative metrics to assess a potential for adoption of this technology. Fig. 10 illustrates price ratio versus the degree days for the selected cities. The ideal case for the installation of the PC systems is when the price ratio and degree days are both high. A higher price ratio shows a significant variation between the base and peak electricity price. When the difference between the base and peak price is not significant, the price ratio is close to zero, suggesting that there is no inherent incentive from the utility programs to deploy the PC systems. Consequently, a suggestion is to first consider cities with price ratios close to one. Another group of candidates for installation of the PC system are the cities with high cooling degree days, suggesting a higher difference between the indoor and outdoor temperatures that serves as a proxy for the cooling load. These cities have high cooling demands, so the load shift would produce noticeable savings in electric utility bills. The results in Fig. 10 indicate that four distinct clusters exist as following: (1) Cities with low price ratio and low degree days, (2) Cities with high price ratio and low degree days, (3) Cities with low price ratio and high degree days, and (4) Cities with high price ratio and high degree days. San Francisco and Chicago are examples of the first cluster, and these cities do not show promising results for the deployment of these PC systems for the cooling season. These cities fall into the group of cities that do not require a high cooling demand and the utility programs do not offer a high price ratio. Consequently, a deployment of the PC systems for cities similar to San Francisco and Chicago during the cooling season requires

careful consideration. New York City and Minneapolis are examples of the second cluster. For example, the results show that for these cities with price ratio and CDDs higher than 0.6 and 200
C, respectively, are among the cities with the potential for the installation of PC systems. Honolulu is an example of the third cluster that although the price ratio is not high, but due to the high cooling demand, the city is one of the promising choices for the installation of the PC systems. The practical choices for the deployment of the PC systems are the cities with high price ratios and CDDs similar to the results of this study for Austin and Phoenix. Overall, for the selected cities in this study, a price ratio more than 0.6 and CDD higher than 200
C are good indicators for the use of PC systems. Based on the identified clusters, it is clear that among the reviewed cities, Phoenix has the highest potential cash savings due to the high CDD and significant price variation. Overall, Phoenix, Austin, Minneapolis, New York City, and Honolulu meet the selection criteria and have promising implication for PC systems, while Chicago and San Francisco would not be promising locations for the studied PC system. 6. Extrapolation to the U.S. building stock To extend the results of this study to the U.S. building stock, this study uses the RECS database for the multi-family residential buildings. The extrapolation of the results requires consideration of various assumptions on the location of the buildings in terms of

502

M. Heidarinejad et al. / Journal of Cleaner Production 171 (2018) 491e505

Fig. 11. Potential energy savings from the deployment of PCs in the residential buildings.

climate and utility rates. This study considers two different scenarios that all residential buildings have similar potential energy saving as (1) the old midrise apartments and (2) the new midrise apartments reviewed in this study based on their climate zones. The selection of these scenarios allows for identifying the upper and lowering saving limits with consideration of old and new apartments, respectively. Fig. 11 illustrates the potential energy savings based on the two scenarios. The potential energy savings are 5.2% and 3.3% for Scenario 1 and Scenario 2, respectively. These potential savings correspond to 0.41 EJ (0.39 Quad) and 0.23 EJ (0.22 Quad) of energy

for Scenario 1 and Scenario 2, respectively. The assumptions include that PCs are located in the space for the midrise apartments. RECS database categorizes the U.S. homes in five different climate zones; this assumption is also part of this extrapolation. Consequently, the potential energy savings for the deployment of PCs in the residential buildings vary from 3.3% to 5.2%. 7. Greenhouse gas emission assessment Fig. 12 provides the CO2 emissions reduction during cooling seasons for old and new office buildings located in the selected

Fig. 12. Office CO2 emissions reduction: (a) Old (b) New buildings.

M. Heidarinejad et al. / Journal of Cleaner Production 171 (2018) 491e505

503

Fig. 13. Midrise apartment CO2 emissions reduction: (a) Old (b) New buildings.

cities. A comparison between the old and new offices indicates a higher CO2 emissions savings from the deployment of portable condoning systems in the old offices. The extended setpoint temperature in the summer ranges from 2.3% to 5.2% and 3.0%e7.1% for old and new offices in the U.S. This savings varies from 0.4% to 2.9% and 2.6% to 3.5% for the PCs deployment. In terms of the CO2 emissions reduction the offices are not the best case studies. Fig. 13 provides the potential CO2 emissions reduction from the deployment of PCs in the midrise apartments. A critical review of the results depicted in Figs. 12 and 13 show that the midrise apartment buildings are better building types than the office buildings for the deployment of PCs. Similar to the offices, the old buildings tend to show better projected CO2 emissions reduction compared to the new buildings. The savings for the extended setpoint is up to 21.8% for the old midrise apartment buildings while this saving is up to 16.5% for the new midrise apartment buildings. For the deployment of the PCs, the CO2 emissions reduction varies from 0.3% to 9.4% and 0.5%e7.8% for old and new midrise apartments, respectively. These projected savings are only due to the building-site CO2 emissions reduction associated with the electricity consumption savings and not the potential CO2 emissions reduction from decreasing the peak demand on the power plants. 8. Discussions 8.1. Broader implications The results of this study also have indirect implications on energy and cost savings. Although the results of this study indicate the benefits of using PCs systems is in cost savings rather than energy savings, indirectly this study could serve as an opportunity to curb energy and greenhouse gas emissions. Due to the cost saving incentives, building managers of commercial office buildings and residents of multi-family building tend to shift their building peak electricity consumption from the peak hours to mid-peak or offpeak hours. This shift will ultimately lead to national wide reduction in the energy and cost savings since the peak shifting ensure power plants are working on (1) their design operation mode rather than their part-load operation mode and (2) full capacity composed to not using the full capacity. Peak shifting and providing

a flat electricity load enable higher marginal cost and increase the baseline load of the power plants that have lower marginal cost (Qiu et al., 2016). Another opportunity for the building portfolios with on-site electricity generation is to promote the use of PC systems during the peak electricity demand. For example, when the campuses have on-site wind turbines or solar collectors, the campuses could support use of PC during the electricity generation and avoid the need to store or sell the excess electricity generated onsite. Overall, besides the first order calculation on the building sites, there are intrinsic and extrinsic benefits from the deployment of PCs during cooling seasons to reduce city energy footprints. Integration of the PC systems with the centralized system could also offer adaptive thermal comfort strategies. Studies showed that the occupants' thermal history (Fadeyi, 2014) and thermal sensation (Chen et al., 2016) render to inefficient energy consumption of the buildings and indoor thermal comfort. Thus, PC systems could serve to address complaint from the unsatisfied occupants. For example, building managers could use PC cooling systems when the occupants enter the building and indicates a warm thermal sensation based on the occupants’ thermal history. Overall, the PC systems provide opportunities for the adaptive thermal comfort strategies. The results of this study could support the existing work to reduce the GHG emissions from the electric sector. Currently, the electric sector is responsible for 41% of the world CO2 emissions (Johnston and Wilson, 2012). Hence, any decarbonizing strategy requires consideration of the electric sector. The results of this study for the eligible cities, e.g. Austin or Honolulu, not only lead to a direct reduction in the electricity consumption, but also it has a direct impact on the annual peak demand. In 2013, the electricity grid in the U.S. required 966 GW while the 786 GW is to address the annual peak demand and 117 GW reserve supply (U.S. Energy Information Administration (EIA), 2013; 2012 long-term reliability assessment, 2012). Therefore, any strategy that address the annual peak demand would have impact on the GHG emission reduction. 8.2. Additional considerations Another possibility for the deployment of PC systems is for the

504

M. Heidarinejad et al. / Journal of Cleaner Production 171 (2018) 491e505

Fig. 14. LMP and TOU for Maryland (a) averaged daily LMP cost versus TOU and (b) averaged hourly LMP cost versus TOU.

heating mode during cold days. Current building design practices suggest using more efficient energy commodities, e.g. natural gas, rather than the electricity. Centralized HVAC system usually relies on natural gas. Currently, there is no detailed plan for the residential and commercial buildings TOU for natural gas. Hence, this study cannot provide a fair energy, cost, and GHG emission assessments for the heating mode of these PC systems. The large deployment of portable PC systems also requires understanding the cost for the Building Automation System (BAS). The new construction buildings usually have BAS with the ability to provide computerized operation schedules, e.g. temperature setpoints. Consequently, these new construction buildings could benefit from the proposed strategy in this study to utilize the PC systems during the peak hours. However, old buildings typically do not have BAS that allow automatic variation of the temperature setpoints. Therefore, the deployment of the proposed PC systems needs to consider the capital cost for the replacement of BAS for the old office buildings without BAS that could lead to up to $100,000 for a medium-sized office building. Beyond the TOU, utility providers offer LMP for large consumers such as the university campuses. The use of TOU allows a long-term prediction for the deployment of PC systems since the cost for the utility is available in advance. However, the LMP requires a continuous prediction due to the dynamic variation of utility price during each hour. Depending on the LMP program, the time window could vary from ½ hours to 2 h to predict future load and cost for the next day. Additional complexity emerges when there are various utility purchase programs such as block purchases or demand response. Utility provides are moving toward shorter time periods for the demand response purchase shorter than the 2 h. Consequently, factors besides the TOU could also affect the usage of PC systems. Fig. 14 provides an example between comparing the LMP and TOU data for Maryland. The TOU for Maryland based on the local provider shows less variation during the peak hours to the offhours. The TOU option for Maryland shows the electricity generation rates of 8.017 cents/kWh, 7.933 cents/kWh, and 7.263 cents/ kWh for the peak hours, mid-peak, and off-peak hours, respectively (PEPCO electricity rate schedules, 2017). This makes the deployment of PCs less practical. However, due to the existence of programs such as LMP, the PCs could be a practical approach to reduce electric utility costs. Fig. 14(a) shows that the average daily prices could vary up to three times. Fig. 14(b) also shows up to three times higher electricity rates for the peak hours compared to the base rates during the off-peak hours. The results of this variation confirm a continuous communication between the centralized systems and

PCs could provide additional opportunities to save energy and reduce GHG emissions. The benefit of signing up for the LMP is that the utility providers could enable running the power plants at the peak design and do not operate the plants in part loads that could lead to excess electricity generation and potential additional GHG emissions. 9. Conclusions This study assessed the impact of deploying Personalized Conditioning (PC) systems supported with a Phase Change Material (PCM) for new and old office and midrise apartment buildings. The aim of this study was to calculate potential energy and cost savings for four building types and seven climate zones in the U.S. The proposed scenarios considered the savings while the buildings operated within standard comfort conditions, extended thermal comfort conditions, and the extended thermal comfort conditions with the consideration of PCs to address any dissatisfied occupants. The results of this study showed a strong direct impact on the cost savings and indirect impacts on the energy savings and CO2 emissions that have implications for the reduction of city energy footprint. A practical approach to footprint reduction utilizes these PCs during the peak hours when the price of electricity is high, and regenerates the PCM during the off-peak hours when the price of electricity is low. There are two options to reject the collected heat during the on-peak time: (1) inside heat rejection, and (2) outside heat rejection. The inside heat rejection could reduce the potential energy savings by about 8%e36% depending on the building's type and location. Midrise apartments demonstrated the most annual savings per person with as $130 per person (or $260 per PC) for the city of Honolulu. Cities with low cooling demand and small variation in the utility rates between the off-peak and on-peak, including Chicago and San Francisco have negligible electricity cost savings. Overall, the cooling energy savings range from 10% t0 70% as compared to the baseline energy consumption. The results of this study shows that operating buildings at 26.7
C has the most potential energy and cost savings for cities with Cooling Degree Days (CDD) of 18
C higher than 300 and TOU variation of 0.5 or higher. In addition, with the placement of PCs, the associated energy penalty could be up to 7% due to the increase in the internal equipment energy. The cost of the PCM based PC systems and local electricity prices are the direct reasons for the actual adaptation of the PCM based PC systems. Indirect adaptation of these PC systems would consider opportunities to curb the energy and greenhouse gas emissions because of the shift in the peak-hours to off-peak hours and increase of the power plant efficiency.

M. Heidarinejad et al. / Journal of Cleaner Production 171 (2018) 491e505

Acknowledgement This study was supported by the project "Robotic Personal Conditioning Device (RoCo)" (05/15e05/18), sponsored by DOE ARPA-E (DEAR0000530). The authors are grateful to Center for Environmental Energy Engineering (CEEE) colleagues at the University of Maryland for providing technical details of a prototype personalize conditioning device, RoCo, during the course of this study. Appendix A. Supplementary data Supplementary data related to this article can be found at https://doi.org/10.1016/j.jclepro.2017.10.008. References 2010 Buildings Energy Data Book, 2011. Table 3.2.7. U.S. Department of energy. 2012 long-term reliability assessment 2012. Attari, S.Z., DeKay, M.L., Davidson, C.I., Bruine de Bruin, W., 2010. Public perceptions of energy consumption and savings. Proc. Natl. Acad. Sci. 107 (37), 16054e16059. Belzer, D., 2014. A Comprehensive System of Energy Intensity Indicators for the U.S.: Methods, Data and Key Trends Pacific Northwest National Laboratory. Brager, G., Zhang, H., Arens, E., 2015. Evolving opportunities for providing thermal comfort. Build. Res. Inf. 43 (3), 274e287. Chang, M.-C., 2015. Room for improvement in low carbon economies of G7 and BRICS countries based on the analysis of energy efficiency and environmental Kuznets curves. J. Clean. Prod. 99, 140e151. Chen, X., Wang, Q., Srebric, J., 2016. Occupant feedback based model predictive control for thermal comfort and energy optimization: a chamber experimental evaluation. Appl. Energy 164, 341e351. City of Austin electric tariff. 2016. City of Austin residential utility rates. 2016.http://austinenergy.com/wps/wcm/ connect/ab6d045c-643e-4c16-921f-c76fa0fee2bf/ FY2016aeElectricRateSchedule.pdf?MOD¼AJPERES. Commercial Building Energy Consumption Survey (CBECS), 2012. U.S. Energy Information Administration (EIA). http://www.eia.gov/consumption/commercial. Dahlhausen, M., Heidarinejad, M., Srebric, J., 2015. Building energy retrofits under capital constraints and greenhouse gas pricing scenarios. Energ Build. 107, 407e416. Davis, S.J., Caldeira, K., Matthews, H.D., 2010. Future CO2 emissions and climate change from existing energy infrastructure. Science 329 (5997), 1330. Deru, M., Field, K., Studer, D., Benne, K., Griffith, B., Torcellini, P., Liu, B., Halverson, M., Winiarski, D., Yazdanian, M., Huang, J., Crawley, D., 2011. U.S. Department of Energy Commercial Reference Building Models of the National Building Stock. NREL. Dhumane, R., Ling, J., Aute, V., Radermacher, R., Mikkilineni, A., Bingham, P., 2016. Transient multiphysics modeling of a robotic personal air-conditioning device. In: 16th International Refrigeration and Air Conditioning Conference. Purdue, 2302e2301 to 2301-2310. Dhumane, R., Du, Y., Ling, J., Aute, V., Radermacher, R., 2016. Transient modeling of a thermosiphon based air conditioner with compact thermal storage: modeling and validation. In: 16th International Refrigeration and Air Conditioning Conference. Purdue, 2144e2141 to 2144-2110. Du, Y., 2016. Battery Powered Portable Vapor Compression Cycle System with PCM Condenser. Mechanical Engineering, University of Maryland. Fadeyi, M.O., 2014. Initial study on the impact of thermal history on building occupants' thermal assessments in actual air-conditioned office buildings. Build. Environ. 80 (0), 36e47. Hannah, H., 2015. How cities can beat the heat. Nature 524, 402e404. Hawaii commerical utility rates, (2016). Hawaii residential utility rates. 2016.https://www.hawaiianelectric.com/saveenergy-and-money/time-of-use-program. Heidarinejad, M., 2014. Relative Significance of Heat Transfer Processes to Quantify Tradeoffs between Complexity and Accuracy of Energy Simulations with a Building Energy Use Patterns Classification. Mechanical Engineering, Pennsylvania State University. Heidarinejad, M., Dahlhausen, M., McMahon, S., Pyke, C., Srebric, J., 2014. Cluster

505

analysis of simulated energy use for LEED certified U.S. office buildings. Energ Build. 85 (0), 86e97. ~ o-Laurent, J.G., Wentz, J.R., Rekstad, N.M., Spengler, J.D., Heidarinejad, M., Ceden Srebric, J., 2017. Actual building energy use patterns and their implications for predictive modeling. Energy Convers. Manag. 144, 164e180. Hoyt, T., Arens, E., Zhang, H., 2015. Extending air temperature setpoints: simulated energy savings and design considerations for new and retrofit buildings. Build. Environ. 88, 89e96. Huang, J., Gurney, K.R., 2016. The variation of climate change impact on building energy consumption to building type and spatiotemporal scale. Energy 111, 137e153. Jessoe, K., Rapson, D., 2015. Commercial and industrial demand response under mandatory time-of-use electricity pricing. J. Ind. Econ. 63 (3), 397e421. Johnston, L., Wilson, R., 2012. Global Power Best Practice Series: Strategies for Fecarbonizing the Electric Power Supply. Synapse Energy Economics, Inc. Kim, Y.-S., Heidarinejad, M., Dahlhausen, M., Srebric, J., 2017. Building energy model calibration with schedules derived from electricity use data. Appl. Energy 190, 997e1007. Kirkeide, L., 2012. Effects of three-hour on-peak time-of-use plan on residential demand during hot Phoenix summers. Electr. J. 25 (4), 48e62. Liu, J., Heidarinejad, M., Gracik, S., Srebric, J., Yu, N., 2015. An indirect validation of convective heat transfer coefficients (CHTCs) for external building surfaces in an actual urban environment. Build. Simul. 8 (3), 337e352. J. Meng, Z. Mi, H. Yang, Y. Shan, D. Guan, J. Liu, The consumption-based black carbon emissions of China's megacities, J. Clean. Prod.. Mi, Z., Wei, Y.-M., Wang, B., Meng, J., Liu, Z., Shan, Y., Liu, J., Guan, D., 2017. Socioeconomic impact assessment of China's CO2 emissions peak prior to 2030. J. Clean. Prod. 142, 2227e2236. Part 4. Minnapolis commercial utility rates. 2016.https://www.xcelenergy.com/staticfiles/ xe/PDF/Marketing/MN-SST-Interim-Rates.pdf. Minnapolis residential utility rates. 2016.https://www.xcelenergy.com/staticfiles/ xe/Regulatory/Regulatory%20PDFs/rates/MN/MNResRateCard.pdf. Moura, M.C.P., Smith, S.J., Belzer, D.B., 2015. 120 Years of U.S. Residential housing stock and floor space. PLOS One 10 (8) e0134135. New York City commercial utility rates. 2016.http://www.coned.com/ customercentral/energyresvoluntary.asp. New York City residential utility rates. 2016.http://www.coned.com/ customercentral/calculators/voluntary_time_of_use.html. PEPCO electricity rate schedules. 2017.http://www.pepco.com/Rate-Schedules.aspx. Phoenix commercial utility rates. 2016.http://www.srpnet.com/prices/business/tou. aspx. Phoenix residential utility rates. 2016.https://www.aps.com/en/residential/ accountservices/serviceplans/Pages/time-advantage.aspx. Qiu, Y., Kirkeide, L., Wang, Y.D., 2016. Effects of Voluntary Time-of-use Pricing on Summer Electricity Usage of Business Customers. Environmental and Resource Economics, pp. 1e24. Raman, A.P., Anoma, M.A., Zhu, L., Rephaeli, E., Fan, S., 2014. Passive radiative cooling below ambient air temperature under direct sunlight. Nature 515 (7528), 540e544. San Francisco commercial utility rates. 2016.https://www.pge.com/en_US/business/ rate-plans/rate-plans/time-of-use/time-of-use.page?. San Francisco residential utility rates. 2016.https://www.pge.com/en_US/ residential/rate-plans/rate-plan-options/time-of-use-base-plan/time-of-useplan.page?. Shen, P., 2017. Impacts of climate change on U.S. building energy use by using downscaled hourly future weather data. Energ Build. 134, 61e70. Shepherd, T.G., 2015. Climate science: the dynamics of temperature extremes. Nature 522 (7557), 425e427. ASHRAE Standard 169-2006, 2006. Weather Data for Building Design Standards. American Society of Heating, Refrigerating, and air-conditioning Engineers, Inc., Atlanta, GA. U.S. Energy Information Administration (EIA), 2013. How Much Electric Supply Capacity Is Needed to Keep U.S. Electricity Grids Reliable? https://www.eia.gov/ todayinenergy/detail.php?id¼9671#. Van Hoof, J., 2008. Forty years of Fanger's model of thermal comfort: comfort for all? Indoor Air 18 (3), 182e201. Veselý, M., Zeiler, W., 2014. Personalized conditioning and its impact on thermal comfort and energy performance e A review. Renew. Sustain. Energy Rev. 34, 401e408. Zhang, H., Arens, E., Taub, M., Dickerhoff, D., Bauman, F., Fountain, M., Pasut, W., Fannon, D., Zhai, Y., Pigman, M., 2015. Using footwarmers in offices for thermal comfort and energy savings. Energ Build. 104, 233e243. Zhu, S., Dalgo, D., Srebric, J., Kato, S., 2017. Cooling efficiency of a spot-type personalized air-conditioner. Build. Environ. 121, 35e48.