research ourna - perkins and will

research journal

ww

w.perkinsw

ill.com

2018 / VOL 10.01

7

01.Energy-Related Occupant Behavior Schedule: Rethinking the Occupant Schedule in Energy Simulations Cheney Chen, Ph.D., P.Eng., BEMP, CPHD, LEED AP BD+C [email protected]

ABSTRACTThis research article focuses on one of the potential gaps in energy simulations, static occupancy schedules. Static occupancy schedules bring uncertainties to energy models and impact the accuracy of energy modeling outputs. In order to generate reliable dynamic schedules, a new workflow is proposed that integrates measured occupant behaviors and improved occupancy schedules. This workflow was tested through occupancy measure-ments, data processing, schedule customization, energy model integration, and eventually energy modeling of a case study building, a commercial office space. The results indicate that measured occupancy schedules result in more accurate energy modeling results. The concluding remarks present multiple and flexible approaches to cope with the occupancy schedules and energy modeling requirements at different design stages.

KEYWORDS: occupant schedule, energy models, workflow, static, dynamic

Energy-Related Occupant Behavior Schedule

1.0 INTRODUCTIONThe increasing environmental concerns regarding worldwide excessive energy use and global warming are the driving forces towards more sustainable use of the diminishing natural resources, especially fossil fuels. As one of the major energy consumers, buildings are as-sociated with 21 percent of the world’s delivered energy consumption, which is necessary to fulfil occupants’ requirements1,2. Therefore, it is critical to understand energy use breakdowns in buildings and promote en-ergy efficient design strategies. Energy modeling is an essential approach to fulfil this purpose3. It has been widely used to compute building energy performances throughout the design process in order to make better-informed design decisions4. However, energy models may produce inaccurate results (Figure 1). Among a group of surveyed LEED certified buildings, there is a wide gap between energy use predicted by LEED energy models and the actual building energy perfor-mance5. There are many possible reasons associated

with the discrepancies6. Other than the limitations em-bedded in the algorithms of the simulation tools and modeling strategies, energy simulations may rely on some inputs or assumptions that are not well measured or verified in the first place7,8.

Occupant behavior is considered as one of the deter-mining variables that impact building energy use, but typically is not well represented in the context of energy simulations. According to Nguyen and Aiello’s survey9, occupancy pattern has shown significant impacts on energy performance of HVAC systems, lighting, appli-ances, and building controls. Nearly 60 percent energy use difference could be expected between the opposite occupancy scenarios (conservation behavior vs. waste-ful one). Since HVAC, lighting, and equipment repre-sent up to 85 percent of the total energy use in build-ings10, poorly described occupant behavior in energy simulations certainly brings uncertainties to the energy modeling results11.

RESEARCH JOURNAL / VOL 10.01

8

Due to the significant energy saving potentials, occu-pancy-related research has drawn considerable atten-tion, addressing three major areas. First, researchers attempted to monitor occupancy behavior for improved understanding of the inter-relationships between its profiles and building energy performances12,13,14,15,16,17. Real-time data was collected as valuable information, since it can be used to optimize applicable occupan-cy profiles. Following up with the site measurements, occupancy-related energy performance was also widely explored and verified via energy simulations. Predicted results confirmed tangible energy savings, ranging from 10 to 60 percent18,19. The energy savings are actually

achieved through occupancy-related control (reflected as modified schedules in the simulation programs). Re-cently, more researchers have focused on occupant be-havior simulations20, 21. A research team from Berkeley Labs and Tsinghua University introduced a novel ap-proach for building occupancy simulation. It is based on the Homogeneous Markov chain model to simulate occupants’ stochastic movement process (Figure 2). A web-based occupancy simulator was also developed, which could take high-level input (occupants, spaces and events) for simulating occupant movement and eventually generate occupant schedules for each space defined in the simulator.

Figure 1: Measured versus proposed savings percentages5.

9

Regardless if they are obtained from measurements or simulations, occupancy schedules deserve to be further studied in order to facilitate more accurate energy cal-culations from energy modeling perspective. At present, they are not very well defined in a wide variety of en-ergy simulation tools, including but not limited to ESP-r, TRNSYs, DOE-2, BLAST, EnergyPlus, and IES Virtual Environment. It is very common that simplified default occupant schedules (e.g. ASHRAE occupancy sched-ules) are adopted in these tools for some typical archi-types (Figure 3). They may be reasonable, but far from accurate when occupants are assumed to “behave” in a static manner in energy simulations. The potential

dynamic interactions between occupants and building systems are underestimated.

By understanding the importance of occupant behavior in buildings and the potential gap embedded in the cur-rent simulation tools, the objective of this research was to explore a usable workflow with which energy mod-elers can ensure more accurate simulation outputs by adopting dynamic occupancy schedules. The hypoth-esis is that the improved schedule can help close the performance gap caused by default schedules (Figure 4).

Figure 2: Online occupancy simulator. (Source: http://occupancysimulator.lbl.gov/)



10

Figure 3: Typical ASHRAE occupancy schedules.

Figure 4: Closing the performance gap through different schedules.

11

2.0 METHODOLOGYA new workflow is proposed in Figure 5, which dem-onstrates a complete process of addressing occupancy measurements, data processing, schedule customiza-tion, energy model integration and eventually energy result verification. This approach relies on dynamic oc-

cupancy schedules in order to account for the com-plex web of interactions between human behavior and building system performance. It is also anticipated that energy model is able to produce more accurate results by adopting the workflow.

Figure 5: The proposed workflow for improved energy modeling process (occupancy measurements - data processing - schedule customization - energy model integration - energy result verification).


12


Figure 6: The locations of the occupancy measurements.

2.1 Occupancy Measurements Occupancy measurement is apparently a straight-forward way to record human behavior in buildings. However, it might be time-consuming to capture any irregular human behavior that requires a long period of monitoring. A field measurement was carried out in the Perkins and Will Vancouver office from 19 June to 12 September 2017. Due to the time constraint, five ar-chetypal locations (four space occupancy types) were chosen, and each location was measured for a period of two weeks (Figure 6). As a simple office environment,

it was expected that the occupancy behaviors in the of-fice environment are regular and trackable. The two-week period would be considered enough to capture the regular patterns of human behavior in these spaces. The four typical office occupancy types covered in the measurements were:

• Meeting room (Basement boardroom, 3rd floor meeting room)

• Office workspace (3rd floor staff workstation)• Kitchen (3rd floor kitchen/copy)• Lavatory (3rd floor washroom).

13


Figure 7: Data loggers.

In the measurements, two types of data loggers (Fig-ure 7) were used to measure occupants’ behavior and some environmental variables:

• HOBO Occupancy & Light data logger moni-tors room occupancy and indoor light changes to identify occupancy patterns and determine energy usage and potential savings.

• HOBO Temperature/Relative Humidity (RH) data logger records temperature and relative

humidity (within 3.5 percent accuracy) in in-door environments with its integrated sensors.

Occupancy behavior was the key parameter in this re-search, but lighting (on/off), ambient air temperature, and RH were also measured as valuable references to help analyze occupancy patterns in the rooms.

14


Figure 8: Converting raw data to derived information.

2.2 Data ProcessingWhen all the measurements were completed, the raw data (downloaded directly from the data loggers) did not indicate structured information. In order to under-stand the data, some processing was necessary to con-

vert raw data into useful information. A typical 24-hr schedule at 10-min interval was generated based on the two-week data collection (Figure 8). Data obtained from the weekdays and the weekends were treated sep-arately for each space.

15

Figure 9: Calibrating occupancy simulation by measured typical schedule.


2.3 Schedule CustomizationThe online occupancy simulator takes “high level” input for occupants, spaces and events before any usable oc-cupant schedule for a given space could be generated. Schedule customization (Figure 9) was performed to ensure the measured occupancy behavior, including

arrival/departure/short term leaving and other possible “events”, can be reflected as high quality inputs ad-opted by the online occupancy simulator. Meanwhile, time allocated to the various events could be identified and used for the occupancy simulator with confidence.

16


Figure 10: Manual input versus ERGON.

2.4 Energy Model Integration Although the simulated schedules can be downloaded as .csv or .idf format and used for EnergyPlus simula-tions, they are not directly usable for other energy simu-lation tools, such as IES Virtual Environment. Therefore, the next step required conversion of the simulated schedules into an IES Virtual Environment compatible. Usually occupancy schedules could be generated in simulation tools quickly with limited data inputs (e.g. 24 inputs represent a typical day in a schedule). However,

manual schedule input becomes an unrealistic ap-proach when hundreds or even thousands of data entry points are necessary for a full year dynamic schedule, especially with a very narrow time-step. Therefore, IES-ERGON (Figure 10) was chosen as a cloud service to import .csv files, graphically interrogate the complete-ness of data, undertake some initial analysis using in-build analytics, and export as a Free Form Profile Data file (.ffd) into IES Virtual Environment simulations.

17

Figure 11: Rapid energy modeling of Perkins and Will Vancouver office by using IES-VE.


2.5 Energy Result Verification Finally, a case study was modeled by using IES Virtual Environment in order to test the workflow and its de-liverable, dynamic occupancy schedules. Perkins and Will Vancouver office was modeled twice using rapid energy modeling strategy22. This strategy relies on three simple steps: capture, model and analyze, and is appli-

cable to capturing existing building conditions. Without changing any other modeling parameters or assump-tions adopted by the rapid energy modeling strategy, default occupancy schedules and measured/simulated ones were adopted in before- and an after- energy sim-ulations. The calculated energy results were compared to an energy audit report.

18


0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

0:00

1:00

2:00

3:00

4:00

5:00

6:00

7:00

8:00

9:00

10:0

011

:00

12:0

013

:00

14:0

015

:00

16:0

017

:00

18:0

019

:00

20:0

021

:00

22:0

023

:00

Measured office weekdays

Figure 12: Measured schedule – staff.

3.0 RESULTS AND DISCUSSIONPerkins and Will Vancouver office’s specific occupan-cy behaviors and energy performance are reported through the results obtained from both the measure-ments and the energy simulations.

3.1 Field Measurements and Schedules In an office environment, staff behavior is a critical piece in shaping the office’s occupancy pattern. A typi-cal 24 hr staff schedule is presented in Figure 12. On average, 80 to 90 percent of workstation occupancy during working hours was observed, whereas 10 to 20 percent working hours are dedicated for some off-work-station activities, such as internal or external meetings, site visits, washroom/kitchen visits, etc. During lunch time, staff desk occupancy is dropped down to 50 per-cent. Outside of working hours, 5 to 10 percent over-time was observed until 9 pm (this is highly relevant to

the individual staff behavior). Unfortunately, due to the time constraint, the workstation measurement focused on a single staff only. It would be more representative if the occupant sampling size (i.e. various staff in the of-fice including design staff, project managers, marketing staff) could be increased. It should be noted that the measured staff schedule aligns with ASHREA office occupancy schedule per-fectly (Figure 13). On one hand, it indicates that the ASHRAE default schedule is not an arbitrary assump-tion, but reasonably based on previous measurements or observations. But, ASHRAE default schedules are limited (nine types) and static in terms of defining the dynamic movement of occupants within buildings. In most cases, occupant numbers are overestimated since the diversity of building occupancy cannot be captured when default ASHRAE schedules are applied to each space.

19


0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%0:00

1:00

2:00

3:00

4:00

5:00

6:00

7:00

8:00

9:00

10:00

11:00

12:00

13:00

14:00

15:00

16:00

17:00

18:00

19:00

20:00

21:00

22:00

23:00

ASHRAE Office weekdays

Figure 13: ASHRAE schedule – staff.

The measurements conducted in the two meeting rooms generated schedules that are beyond the scope of typical ASHRAE schedules. They indicate unique meeting occupancy behaviors (Figure 14). The board-room in the basement is extremely popular during morning from 10 am to 12 pm and up to 80 percent occupancy around noon is observed. The reason could be that many formal management meetings and cli-ent meetings are held in this space during this specific time. After the peak hours, the occupancy decreases to around 30 to 40 percent. Usually project teams would take over the space and use it as a project meeting place similar to other meeting rooms in the building. There is still an averaged 10 to 15 percent occupancy from 6 to 9 pm when extended meetings or informal discussions are held occasionally in the basement.

Compared to the large boardroom in the basement, the 3rd floor meeting room represents typical project meet-ing places, which occur at each floor. The occupancy pattern is featured with evenly distributed occupancy in the morning and the afternoon. The peak time is in early morning and late afternoon with around 60 percent oc-

cupancy. Aligning with the staff behavior, the meeting room occupancy is reduced down to 20 percent during lunch hours. Extended meetings are also held in the room after the working hours with less than 10 percent occupancy.

Meeting room schedule is a good example to explain complexities of occupancy pattern within the building. Besides some transient occupants (visitors), the meet-ing rooms are actually occupied by the building’s staff who move from their workstations to the meeting places. The occupancy patterns (both workstation and meeting room) determine the location of the occupants in each time period within the building. Most importantly, when they are located in a space (either workstation or meet-ing space), they will trigger the operation of the build-ing systems, such as lights, HVAC and windows, and impact the energy use of the occupied space as well as the space they left. Without considering the diversity of occupants but simply applying static schedules into each space would ignore such dynamic operation and overestimate energy use.

20


Figure 14: Measured schedule – Conference room: (a) Board room; (b) 3rd floor meeting.

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

0:00

1:00

2:00

3:00

4:00

5:00

6:00

7:00

8:00

9:00

10:0

011

:00

12:0

013

:00

14:0

015

:00

16:0

017

:00

18:0

019

:00

20:0

021

:00

22:0

023

:00

Measured boardrm weekdays

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

0:00

1:00

2:00

3:00

4:00

5:00

6:00

7:00

8:00

9:00

10:0

011

:00

12:0

013

:00

14:0

015

:00

16:0

017

:00

18:0

019

:00

20:0

021

:00

22:0

023

:00

Measured meeting weekdaysMeasured meeting weekdays(b)

(a) Measured boardrm weekdays

21


0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

0:00

1:00

2:00

3:00

4:00

5:00

6:00

7:00

8:00

9:00

10:0

011

:00

12:0

013

:00

14:0

015

:00

16:0

017

:00

18:0

019

:00

20:0

021

:00

22:0

023

:00

Measured kitchen/copy weekdays

Figure 15: Measured schedule – Kitchen.

Kitchen/copy room is not covered by ASHRAE default schedules although it is apparently a popular space vis-ited by the building occupants at each floor. From the measurements, the kitchen is frequently visited from 8 am until 8 pm throughout a day. Afterwards, janitorial contractors and overtime staff’s activities are still ob-served until midnight. It is expected that the kitchen occupancy is not high in each time period compared as workstations or meeting rooms. The reason is that the activities happening in the space are usually short-term, although it serves about 30 occupants on each floor. It is interesting to observe the two peaks (around 50 percent occupancy) at around 8 to 10 am and 12

to 2 pm that echo breakfast- and lunch-related activi-ties. Similar to the workstation and the meeting place, occupants within the kitchen will trigger the operation of kitchen/copy systems, such as water fixtures, micro oven, coffee machine and printers. In order to under-stand kitchen energy use pattern better, the activities reflected in the schedule are translated to an effective use time per day. An average of 220 minutes of kitchen/copy room use per day is calculated. Without sub-me-tering installed in the place, the effective use time per day could be useful for estimating the energy use of kitchen/copy systems.

22


0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

0:00

1:00

2:00

3:00

4:00

5:00

6:00

7:00

8:00

9:00

10:0

011

:00

12:0

013

:00

14:0

015

:00

16:0

017

:00

18:0

019

:00

20:0

021

:00

22:0

023

:00

Measured washrm weekdays

Figure 16: Measured schedule – Lavatory.

On average, 20 to 30 percent occupancy throughout a day can be observed in the measured lavatory. Consid-ering the ratio of users per lavatory per floor at 15/1 (the assumption is 30 staff per floor with 15 males and 15 females), the consistent occupancy is acceptable and explains staff’s behavior other than in the workstations, the meeting rooms, the kitchens and outside of the building. Similar to the kitchen, the activities reflected

in the schedule can be translated into a total of 121 effective minutes of lavatory use per day or 8 minutes per person per day. Leadership in Energy and Environ-mental Design (LEED) assumes default water closet use rate as 3 times a day for building occupants. The ob-servations obtain from the lavatory align with the LEED assumptions and they are useful for estimating water use and lighting energy use in the lavatory.

23


Figure 17: A comparison of REM and Level 2 Energy Audit results (with ASHRAE schedules).

3.2 Before and After Energy SimulationA before and after energy simulation was performed. By following the rapid energy modeling strategy, Per-kins and Will Vancouver office was first modeled with respect to the following aspects:

• ASHRAE default office building schedules (occupant, lighting and equipment)

• Office drawings/layout/dimensions • Building envelope thermal performance esti-

mations by walkthrough • Building HVAC system walkthrough • Building lighting system walkthrough• Building equipment specification sheets• Weather file - Vancouver 718920 (CWEC).

By comparing the rapid energy modelling results and a Level 2 Energy Audit for LEED O&M EAc2.1 report (Fig-ure 17), it is observed that the total energy difference is at 16 percent, which fulfils the initial expectation of the

variance within 20 percent. However, the larger differ-ences, up to 146 percent, are observed when energy breakdowns are compared in detail. Hot water, artificial lighting and equipment energy use are all closely re-lated to the occupancy behaviors in the building. Ap-parently, the ASHRAE default occupancy schedules do not track the interrelationship between occupant and building system energy use correctly, and trigger the discrepancies. On the other hand, HVAC related end-uses, such as fan energy, heating and cooling energy, are also off-track. However, HVAC energy use is much more complicated because these systems are impacted by not only occupancy behavior but also micro-climate, building envelope performance, internal heat gain, infil-tration, natural ventilation strategy, and system perfor-mance. Therefore, HVAC related energy use may not have strong correlation with occupancy behavior and their accuracy is beyond the current research scope.

24


Figure 18: A comparison of REM with calibrated schedules and Level 2 Energy Audit results (with measured schedules).

The energy simulation was conducted again by only adopting the dynamic full-year schedules, calibrated and produced by the measurements and the online oc-cupant simulator. For the three occupancy driven ener-gy uses (hot water, artificial lighting and equipment/plug loads), their accuracy has been greatly improved (Figure 18). The improvement is observed simply because the new schedules can capture the occupant driven energy uses more accurately. However, HVAC related energy

is still far from the baseline and it indicates the energy simulation accuracy is not only affected by occupancy schedules alone. Interestingly, both space heating ener-gy and cooling energy are also improved slightly in after simulation by roughly 10 to 20 percent (-59 percent vs. -34 percent & 76 percent vs. 66 percent). The contribu-tion is from the corrected internal gains, including oc-cupant, lighting and equipment, which are all improved by the dynamic schedules.

25


Figure 19: Schedule generation approaches.

3.3 Further ConsiderationsThe results indicate that the proposed workflow is able to improve the accuracy of energy simulations through introducing detailed occupancy schedules. On the other hand, there are more ways to produce usable occu-pancy schedules for energy modeling use. The potential schedule generation workflows could be simply sum-marized as Default Schedules (DS), Simulated Sched-ules (SS), Measured & Simulated Schedules (MSS) and Measured Schedules (MS) (Figure 19). DS is widely used in the energy modeling world but MS, MSS, and SS are all relatively new ones. By taking different schedule generation strategies, different levels of accuracy could be anticipated. Although MS, MSS or even SS could generate more accurate energy results, the penalty is always associated with the time-consuming process, es-pecially for MS and MSS.

Instead of using one single approach throughout all the design phrases, energy modelers should consider how to use these workflows in a smart way by addressing their pros and cons (Figure 20). DS could still be use-ful at initial design stage when better assumptions or detailed information are not available from design team or client. Meanwhile, the accuracy of early stage energy modeling is not critical, since sensitivity or parametric studies could still be able to drive the design team to the right design decisions. MSS and MS could update ener-gy model by improving its accuracy when more detailed information is to be measured or simulated at CD or DD stages. It alights with the expectation of more accurate energy prediction at these design stages. SS is dedi-cated to measurement and verification modeling when building metering systems are well established and all the detailed information could be measured precisely over a long period of time.

26


MS

MSS SS DS

UncertaintyTime

Figure 20: Pros and Cons of different schedule generation strategies.

4.0 CONCLUSIONSThis research focuses on one of the potential gaps, oc-cupancy schedule, which brings uncertainties to energy models and impacts the accuracy of energy modeling results. A new workflow was therefore proposed and tested through the occupancy measurements, data processing, schedule customization, and energy model integration. A specific case study was used to test the workflow, and this article discusses results in detail.

The current research has the following limitations, which may require further studies:

• Due to the time constraint, only one staff work-station was measured. The behaviors will be slightly different for all employees in the build-ing. Although it would be difficult to monitor each staff member, at least representative oc-cupant groups, such as design staff, market-ing staff, managing staff, should be measured individually in the future.

• The research focuses on the MSS workflow. A quantitative cross-comparison of all the four approaches, MS, MSS, SS and DS, would be helpful to guide when and how each workflow could be employed in energy modeling for ap-propriate design assistance.

• Not only human behavior, but all the measur-able variables, such as water flow, daylighting and lighting, microclimate, etc. could be inte-grated into the similar workflow for more accu-rate energy simulations. This approach would work well with the new LEED v4 metering re-quirements, which encourages extensive data collection in buildings.

AcknowledgementsThe author would like to express his gratitude to Perkins and Will for the generous research funding with which his 2017 Innovation Incubator project has been com-pleted successfully. The gratitude also extends the man-agement team of Perkins and Will Vancouver office for their continuous support. Final acknowledgement is due to Kathy Wardle for her kind help in refining this paper.

REFERENCES[1] National Resource Canada, (2016). Energy Efficien-cy Trends in Canada, Report, Retrieved on 3/2018 from https://www.nrcan.gc.ca/sites/www.nrcan.gc.ca/files/energy/pdf/trends2013.pdf.

27


[2] U.S. Energy Information Administration, (2017). International Energy Outlook 2017, Report, Retrieved on 3/2018 from https://www.eia.gov/outlooks/ieo/pdf/0484(2017).pdf.

[3] Han, X., Pei, J., Liu, J., and Xu, L., (2013). “Multi-Objective Building Energy Consumption Prediction and Optimization for Eco-Community Planning”, Energy and Buildings, Vol. 66, pp. 22–32.

[4] Hoes, P., Hensenb, L., Loomans, M., de Vries, B., and Bourgeois, D., (2009). “User Behavior in Whole Building Simulation”, Energy and Buildings, Vol. 41, pp. 295-302.

[5] New Buildings Institute, (2008). Energy Perfor-mance of LEED for New Construction Buildings, Report, Retrieved on 3/2018 from https://newbuildings.org/resource/energy-performance-leed-new-construction-buildings/.

[6] Holladay, M., (2012). Musings of and Energy Nerd: Contemplating Residential Energy Use, Retrieved on 3/2018 from http://www.greenbuildingadvisor.com/blogs/dept/musings/energy-modeling-isn-t-very-accu-rate.

[7] Meneze, A., Cripps, A., Bouchlaghem, D., and Buswell, R., (2012). “Predicted vs. Actual Energy Per-formance of Non-Domestic Buildings: Using Post-Oc-cupancy Evaluation Data to Reduce the Performance Gap”, Applied Energy, Vol. 97, pp. 355–364.

[8] Bordass, B., Cohen, R., Standeven, M., and Lea-man, A., (2001). “Assessing Building Performance in Use 3: Energy Performance of Probe Buildings”, Build-ing Research and Information, Vol. 29, pp. 114–28.

[9] Nguyen, T., and Aiello, M. (2013). “Energy Intel-ligent Buildings based on User Activity: A Survey”, En-ergy and Buildings, Vol. 56, pp. 244-257.

[10] National Resource Canada, (2011). 15th Edition of Energy Efficiency Trends in Canada, Report, Retrieved on 3/2018 from www.nrcan.gc.ca.

[11] Clevenger, C., and Haymaker, J., (2006). “The Im-pact of the Building Occupant on Energy Modeling Sim-ulations”, Conference Proceedings - Joint International Conference on Computing and Decision Making in Civil and Building Engineering, Montreal, Canada.

[12] Ueno, T., Sano, F., Saeki, O., and Tsuji, K., (2006). “Effectiveness of an Energy-Consumption Information

System on Energy Savings in Residential Houses Based on Monitored Data”, Applied Energy, Vol. 83, pp. 166–183.

[13] Petersen, J., Shunturov, V., Janda, K., Platt, G., and Weinberger, K., (2007). “Dormitory Residents Re-duce Electricity Consumption When Exposed to Real-Time Visual Feedback and Incentives”, International Journal of Sustainability in Higher Education, Vol. 8, No. 1, pp. 16–33.

[14] Allcott, H., (2010). “Behavior and Energy Policy”, Science, Vol. 327, pp. 1204–1205.

[15] Faruqui, A., Sergici, S., and Sharif, A., (2010). “The Impact of Informational Feedback on Energy Con-sumption—A Survey of the Experimental Evidence”, Energy, Vol. 35, No. 4, pp. 1598-1608.

[16] Peschiera, G., and Taylor, J., (2012). “The Impact of Peer Network Position on Electricity Consumption in Building Occupant Networks Utilizing Energy Feedback Systems,” Energy and Buildings, Vol. 49, pp. 584-590.

[17] Vassileva, I., Odlare, M., Wallin, F., and Dahlquist, E., (2012). “The Impact of Consumers’ Feedback Pref-erences on Domestic Electricity Consumption”, Applied Energy, Vol. 39, pp. 575–582.

[18] Li, N., and Becerik-Gerber, B., (2011). “Perfor-mance-Based Evaluation of RFID-Based Indoor Lo-cation Sensing Solutions for the Built Environment”, Advanced Engineering Informatics, Vol. 25, No. 3, pp. 535–546.

[19] Masoudifar, Na., Hammad, A., and Rezaee, M., (2014). “Monitoring Occupancy and Office Equipment Energy Consumption Using Real-Time Location System and Wireless Energy Meters”, Proceedings of the Winter Simulation Conference, pp. 1108-1119

[20] Feng, X., Yan, D., and Hong, T., (2015). “Simula-tion of Occupancy in Buildings”, Energy and Buildings, Vol. 87, pp. 348-359.

[21] Wang, C., Yan, D., and Jiang, Y., (2011). “A Novel Approach for Building Occupancy Simulation”, Build-ing Simulation, Vol. 4, No. 2, pp. 149-167.

[22] Autodesk, (2011). Streamlining Energy Analysis of Existing Buildings with Rapid Energy Modeling, Re-trieved on 3/2018 from http://images.autodesk.com/adsk/files/rem_white_paper_2011.pdf.

research ourna - perkins and will

Documents