University of Southern California Dissertations and Theses

Real-time simulation-based feedback on carbon impacts for user-engaged temperature management

(USC Thesis Other)

Real-time simulation-based feedback on carbon impacts for user-engaged temperature management

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content i

Real-Time Simulation-Based Feedback on
Carbon Impacts for User-engaged Temperature Management

By
Danyang Zhang
Presented to the
FACULTY OF THE
SCHOOL OF ARCHITECTURE
UNIVERSITY OF SOUTHERN CALIFORNIA
In partial fulfillment of the
Requirements of degree
MASTER OF BUILDING SCIENCE

August 2019

ii

ACKNOWLEDGEMENTS

I would like to express the deepest appreciation to my committee chair, Professor Kyle Konis, for his conscientious
guidance and constant encouragement to accomplish this thesis.

My appreciation also extends to Professor Marc Schiler and Professor Douglas Noble for their valuable guidance and
support for completion of this thesis.

I would also like to thank all of my friends in MBS, who directly or indirectly help and encourage me to complete this
thesis.
At last but not least gratitude goes to my parents and my best friend, Weiqi Chen, for providing me with unfailing
support and continuous encouragement throughout my life. This accomplishment would not have been possible
without them.

Thank you.

iii

COMMITTEE MEMBERS
Chair: Prof. Kyle Konis, Ph.D, AIA
Title: Assistant Professor
Email: kkonis@usc.edu

Second Committee Member: Prof. Marc Schiler
Title: Professor
Email: marcs@usc.edu

Third Committee Member: Prof. Douglas Noble
Title: Associate Professor
Email: dnoble@usc.edu

iv

ABSTRACT
The current control methods of HVAC systems depend on static temperature set points based on standards such as
ASHARE Standard 55, rather than the comfort requriements of the actual building occupants in a space, often leading
to unnecessary over-heating or over-cooling. These control methods in turn result in unnecessary energy consumption
and contribute to higher energy costs and carbon emissions. Although some researchers demonstrated that occupant-
participatory temperature adjustment approach, i.e., incorporating users’ thermal comfort feedback via mobile app or
web, can reduce energy consumption, those studies do not place comfort feedback and associated thermal changes in
context with changes in carbon emissions.
The aim of this thesis is to develop a simulation-based approach to generate the necessary building performance data
to support the development of a carbon-aware virtual thermostat for mobile devices to enable building occupants to
better-understand the impact of temperature setpoint changes on building energy use and associated carbon emissions.
TrojanSense is a project encouraging students, faculty and staff to make changes to the indoor set point temperature
in University of Southern California (USC) buildings. Watt Hall, as the main building housing the architecture school
at USC, was chosen as the sample building. The building model was generated in Rhino and Grasshopper, then setting
up all the parameters and running energy simulations in Honeybee and EnergyPlus. The simulation output is the data
exchange format with all possible set point temperature combinations. By exporting this pre-calculated simulation
results into TrojanSense mobile app or web, the occupant can look up how much energy consumption and carbon
emission could be saved or wasted in real time when virtually adjusting the setpoint, which raises occupants’ carbon
awareness and encourage them to pay more attention to sustainability.
KEY WORDS: Carbon emission, Thermal Comfort, Building Energy Modeling,
HYPOTHESIS
After integrating pre-calculated energy simulation outcomes based on different temperature set points, a real-time
carbon-aware thermostat could be developed to enable users to become aware of the change in carbon emissions for
a given setpoint change.
GOALS
- Raise carbon awareness of student, faculty and staff in USC.
RESEARCH OBJECTIVES
- Simulate energy consumption of the sample building associated with all realistically possible set point combinations.
- Generate a simulation dataset in exchangeable format for utilization by mobile app or web.
- Investigate the approach for campus building energy simulation as the preparation for future work.

v

TABLE OF CONTENTS
ACKNOWLEDGEMENTS .................................................................................................................................... ii COMMITTEE MEMBERS ................................................................................................................................... iii ABSTRACT ......................................................................................................................................................... iv 1.INTRODUCTION ............................................................................................................................................... 1 1.1 USC Sustainability Strategy 2030 ................................................................................................................. 2 1.2 Participatory Sensing: The TrojanSense Project ............................................................................................ 2 1.2.1 Temperature Set Points .......................................................................................................................... 4 1.3 Thermostat and Virtual Thermostat ............................................................................................................... 4 1.4 Environmental/Carbon Aware Thermostat .................................................................................................... 5 1.5 Model Predictive Control (MPC) .................................................................................................................. 5 1.6 Building Archetypes and Campus Building Modeling ................................................................................... 6 1.7 Software ....................................................................................................................................................... 6 1.7.1 Rhino and Grasshopper .......................................................................................................................... 6 1.7.2 Grasshopper plug-in: Meerkat GIS and Honeybee .................................................................................. 6 1.7.3 EnergyPlus and Openstudio ................................................................................................................... 6 1.8 Summary ...................................................................................................................................................... 6 2. BACKGROUND AND LITERATURE REVIEW ............................................................................................... 8 2.1 Thermal Management Approaches in Other Universities ............................................................................... 8 2.1.1 Eastern Mennonite University: A Green Nerd in FM .............................................................................. 8 2.1.2 Wesleyan University: Creating a Successful Campus Temperature Policy .............................................. 8 2.1.3 University of Maryland: Occupant Engagement and Centralized Controls .............................................. 8 2.1.4 Temperature Policy ............................................................................................................................... 8 2.2 Occupant-Participatory Temperature Adjustment Approach .......................................................................... 9 2.2.1 Exaggerated votes................................................................................................................................ 11 2.3 Urban Scale Building Energy Modeling (UBEM) ........................................................................................ 11 2.3.1 Collecting and analyzing of data input ................................................................................................. 11 2.3.2 Generating and simulating of thermal models ....................................................................................... 12 2.4 Summary .................................................................................................................................................... 12 3. METHODOLOGY ........................................................................................................................................... 13 3.1 Goal ........................................................................................................................................................... 13 3.2 Approach ................................................................................................................................................... 14 3.2.1 Data Collection.................................................................................................................................... 15 3.2.2 Energy Simulation ............................................................................................................................... 15
vi

3.2.3 Results Processing and Visualization ................................................................................................... 16 3.3 Static Energy Model (Hand Calculation) ..................................................................................................... 16 3.3.1 Heat Gain by Natural Ventilation ......................................................................................................... 16 3.3.2 Internal Heat Gain ............................................................................................................................... 16 3.3.3 Heat Gain through Envelope (Roof, Walls and Floor) ........................................................................... 16 3.3.4 Grasshopper ........................................................................................................................................ 17 3.4 Dynamic Energy Model .............................................................................................................................. 17 3.4.1 Building Energy Modeling in Honeybee and EnergyPlus...................................................................... 18 3.4.2 Building Energy Modeling: Watt Hall Model ....................................................................................... 19 3.4.3 Building Energy Modeling: Setting up Parameters ............................................................................... 20 3.4.4 EnergyPlus to TrojanSense Data Transfer Format ................................................................................ 24 3.4.5 Background Processing for Carbon-Aware Thermostat......................................................................... 25 3.5 Campus Scale Building Energy Simulation ................................................................................................. 28 3.6 Summary .................................................................................................................................................... 29 4. RESULTS ........................................................................................................................................................ 30 4.1 Weather Data and Room Temperature......................................................................................................... 30 4.2 Building-Level Energy Simulation Results .................................................................................................. 31 4.3 Floor-Level Energy Simulation Results ....................................................................................................... 33 4.5 Zone-Level Energy Simulation Results ....................................................................................................... 38 4.6 Summary .................................................................................................................................................... 43 5. DISCUSSION .................................................................................................................................................. 44 5.1 Analysis of Building Level Energy Simulation Results ................................................................................ 44 5.2 Analysis of Floor Level Energy Simulation Results ..................................................................................... 44 5.3 Analysis of Zone Level Energy Simulation Results ..................................................................................... 46 5.4 Summary .................................................................................................................................................... 50 6. CONCLUSION AND FUTURE WORK ........................................................................................................... 52 6.1 Conclusions ................................................................................................................................................ 52 6.2 Future Work ............................................................................................................................................... 52 6.2.1 Campus Building Energy Modeling ..................................................................................................... 52 6.2.2 Building Retrofit Strategies.................................................................................................................. 52 6.2.3 Validation ........................................................................................................................................... 53 6.2.4 Data Transfer ...................................................................................................................................... 53 6.2.4 Other Improvements ............................................................................................................................ 53 APPENDIX.......................................................................................................................................................... 55
vii

REFERENCE ....................................................................................................................................................... 59

1
1.INTRODUCTION
The active participation of building occupants is becoming more widely used for determining set points of HVAC
system. One of the studies is TrojanSense project, which has started and been promoted in USC campus. However,
for most occupants, they are not aware of carbon impacts when adjusting temperatures. This limitation could lead to
some not environmentally friendly outcomes. As a carbon-aware education tool, the real time energy simulation model
can solve this problem by telling them how much energy consumption and carbon emission will cause or save at each
set points.
The project aims at creating a real-time simulation-based feedback mechanism to tell users of a mobile app what the
carbon impacts will be from a given change (increase / decrease) in the current room temperature. When the user
adjusts the virtual thermostate by 1 degree, the user sees the annual increase or decrease in carbon emissions for that
specific zone that the user is currently in. As figure 1 shows, this study continued the data collection, which already
started and been processed by TrojanSense project. The research mainly focused on energy simulation and results
visualization for users to raise their carbon awareness.

Figure 1: Overall Methodology

In this chapter, the USC sustainability strategy is introduced to understand the general background of this study. The
TrojanSense project, as one of the occupants-participatory temperature adjustment tools, is introduced in detail. The
purpose and significance of having TrojanSense project is explained. Thermostat and how it impacts building energy
use are explained. The definition of the carbon-aware virtual thermostat is introduced and illustrated. The importance
of using dynamic energy simulation instead of only hand calculation for energy consumption is discussed. This chapter
also explains the reason for designing a carbon-aware education tool in USC campus instead of only one single
building. Software related to this research, including Rhino, Grasshopper, Meerkat GIS, Honeybee, EnergyPlus and
Openstudio, are introduced in this chapter.

2

1.1 USC Sustainability Strategy 2030
In 2018, the USC Academic Senate Sustainability Committee developed the USC Sustainability Strategy 2030. The
proposal states that, “USC aims to develop the knowledge, people, and practices needed to lead the way in responding
to the major environmental sustainability challenges facing our city, region, country, and planet.” The current goal
of this strategy consists of seven categories, including education and research, community engagement, energy
conservation, transportation, procurement, waste and water.
The overall proposed goal in energy conservation and greenhouse gas reduction domain is to achieve carbon neutrality
in all campus buildings by 2030. The committee also raised two more specific goals referring to greenhouse gas
emissions and utility costs. The final objective of greenhouse gas domain is to track and report greenhouse gas
emissions, ultimately reducing greenhouse gas emissions per square foot by 20% from 2014 levels to 2020.
1.2 Participatory Sensing: The TrojanSense Project
As the largest energy consumer in Los Angeles, the USC campus is aiming to reach approximately 40% carbon
emission reductions of buildings before 2030 (USC, 2017). On the other hand, traditional control method for HVAC
systems in large buildings is using centralized control strategy of temperature setpoints, which is strictly controlled
by facility management apartment (Angela, Marco and Kiernan, 2016). For most universities, students and faculty are
not able to access the temperature control system. This gap will cause inadequate set points, uncomfortable indoor
environment and unnecessary energy consumption. As the majority of building occupants, students and faculty should
participate in controlling HVAC system (Brager et al, 2015). Moreover, occupant satisfaction will increase when more
students and faculty take part in the temperature adjustment. To support the goal and bridge the gap, a project called
TrojanSense has been developed by students and faculty at USC. By developing this project, users enable to involve
in occupant-engaged temperature management of HVAC system of USC campus. The goal is to reduce energy and
resulting carbon emissions by making setpoint changes that result in reduced need for mechanical HVAC. Currently,
there are eight people in TrojanSense project with different main duties. Professor Kyle Konis is the leader of this
project. As one of the team members, Danyang Zhang is the energy modeler to develop the carbon-aware thermostat.
The TrojanSense includes two user-facing platforms: a TrojanSense web and a mobile application. Both were launched
in Fall 2018. User interaction with web consists of a simple user interface (Figure 2). Users need to submit their
location (building name and room number) firstly. Then a question “Would you like it to be warmer or cooler in here”
is asked to collect user’s feedback. By adjusting the circular and virtual thermostat, users can choose how much degree
(+/- 1 to 6 ºF) they are willing to change in this room. After clicking submit, users can leave comments about this
space. Users can also view the voting summary on the web. In the TrojanSense campus map, users can select any
location and see others voting results in the past (figure 3).

3
Figure 2: TrojanSense Web User Interface (TrojanSense, 2018)

Figure 3: TrojanSense Campus Map and Voting Summary (TrojanSense, 2018)
The interface of mobile application is more convenient than the web since it can detect users’ location automatically
using a beacon, which enables occupants’ smartphones to collect a signal and send a signal back to an online platform.
Before using the TrojanSense mobile application, the user needs to turn on Bluetooth and the smartphone starts
detecting which room the user is currently in. Users can also choose to scan the QR code in the room or enter the room
number manually (Figure 3-a, b). In the next screen, same with web voting, users are asked to choose a temperature
change (Figure 4-d). An opportunity of leaving comments are given to users after they submit their votes (Figure 4-
f).

a b c

4

d e f
Figure 4: TrojanSense Mobile APP User Interface Screens (TrojanSense, 2018)
1.2.1 Temperature Set Points
In temperature control applications, Heating, Ventilation, Air Conditioning (HVAC) system works together to bring
the room temperature to the set point. For example, if the indoor temperature is 75F while set point temperature is
72F, then the air conditioning system must work to decrease the room temperature to the target value. Reaching the
set point temperature is not equal to thermal comfort since thermal comfort are determined by various factors such as
clothing. However, it is possible to optimize set points because of the following reasons: 1) Energy efficient: an
adequate set point temperature can save energy consumption. 2). Durable facilities: product life will be extended when
the operating hours decrease. 3). Increase occupant satisfaction: an optimal set point temperature can satisfy more
occupants’ thermal comfort.
A good strategy for maintaining comfortable indoor temperature range and saving energy consumption is introducing
occupants-engaged temperature management to determine set points. The term, “occupant-engaged” refers to dynamic
temperature set points adjustments informed by data-driven thermal comfort models generated at the thermal zone
level. In occupant-engaged temperature management, occupants provide their acceptable indoor temperature for
specific space to engineers or facility managers. The reported feedbacks will be collected and paired with real-time
temperatures to develop a more humanized comfort model (TrojanSense, 2018).
To implement this strategy, the TrojanSense mobile application is built and distributed for USC students, staff and
faculty. Using a network of Bluetooth beacons distributed at various points on campus, the TrojanSense app can
automatically detect if the user is within range of a beacon, and ask users to report their thermal sensation (e.g. whether
they feel too warm, too cold or neutral) (TrojanSense, 2018). In addition, Internet of Things (loT) sensors are installed
in different space (e.g. library, lecture hall, studio) in USC campus to measure the indoor air temperature and relative
humidity at the same time. Users can report their location and how they would like to adjust the current indoor
temperature, then a thermal comfort model will be built by corresponding occupants’ subjective feedbacks and
measured real-time temperature by loT device. These models will be shared with USC Facilities Management Services
(FMS) contributing to HVAC system set points. The purpose of TrojanSense project is minimizing the amount of
energy consumption and carbon emission caused by overheating and overcooling space.
1.3 Thermostat and Virtual Thermostat

5
A thermostat is a component that automatically controls temperature and regulates temperature reaching a certain set
point. The term thermostat refers to the mechanical device to maintain the demand temperature range by switching
the heating or cooling system on or off. During the heating seasons, when the temperature is below the set point, the
thermostat switches the heater on, then then temperature will increase. When the indoor temperature is higher than the
set point temperature, the heater will be turned off and temperature decrease. When the indoor temperature falls below
the set point temperature, this control cycle will repeat.
Thermostats are used in any device or system that heats or cools to a setpoint temperature, applying in HVAC system
as well as kitchen equipment such as ovens and refrigerators. Thermostat is one of a typical closed-loop control system
example. In a closed loop control system, the feedback is used to determine the adjustment to the drive signal that
change the plant’s state. In thermostat, the heater or the air conditioner is the plant. The thermostat will turn the air
conditioner on when the room temperature is higher than the desired setpoint. The air conditioner will respond to the
signal which come from thermostat continually until the room temperature reach the setpoint. One of the biggest
problems in thermostat control system is the differentiation, which means the plants may not reach the desired output
in a very short time and overshoot or oscillation may happen in the repeat temperature adjustment cycle (Barr, 2001).
In contrast with a “real” physical thermostat in the HVAC control system, the concept of TrojanSense project is using
a “virtual thermostat” to adjust temperature. In the “virtual thermostat”, the loop is still closed, however, the response
is not from a sensor or the thermostat, occupants’ thermal sensation is the virtual sensor in this cycle to control the
response, and the response is no longer immediate but taking a long period.
Occupants play an important role in developing the “virtual thermostat”. When the indoor temperature is too high or
too low than comfortable setpoints temperature, occupants can adjust the temperature to acceptable range using their
“virtual thermostat”-TrojanSense mobile application or website. The virtual thermostat is not an instantaneous
thermostat currently. The current operating principle of virtual thermostat is collecting and organizing data, then
sharing it with Campus Facility Management to adjust setpoint temperature based on postprocessing and analysis of
data collected for a given time period and space.
1.4 Environmental/Carbon Aware Thermostat
For most occupants, the reason for changing setpoint temperature is the indoor temperature making them
uncomfortable. However, the comfortable temperature is a range not just a specific value and is influenced by several
factors, including clothes, gender, height. Occupants are unaware of the impact of changing setpoint temperature and
may vote exaggeratively, the inauthentic votes may have negative effects for optimize set point temperature, which
lead to lower occupants’ thermal comfort and higher energy consumption.
Although virtual thermostat has already made contribution to energy savings, an additional strategy called
environment/carbon aware thermostat can be a further step for energy saving. Compared with virtual thermostat,
carbon aware thermostat not only focus on occupants own thermal sensation, but also helping them understand the
environmental impact of changing setpoint temperature. For instance, in a cooling season, if the user trying to decrease
the current setpoint temperature, the carbon aware thermostat can show how much energy and carbon emissions will
be wasted for that specific zone that the user is currently in. By using carbon aware thermostat, occupants enhanced
their carbon awareness so they will evaluate the importance of energy efficiency and their thermal comfort in next
vote.
1.5 Model Predictive Control (MPC)
Model Predictive Control (MPC) is control strategy that is used to control a process while satisfying a set of constraints.
It is a feedback control algorithm that uses a model to make predictions about future outputs of a process. MPC utilizes
the model of a system to predict its future behavior. It solves an optimization problem at each time step to find the
optimal control action that drives the predicted plant output to the desired reference as close as possible. The
researchers are able to predict future events and changes. Furthermore, the uncertainties of prediction results are
minimized because of the operation in a receding horizon mode. The basic principles of MPC include an internal

6
dynamic model of the process, a history of past control moves and an optimization cost function over the receding
prediction horizon to calculate the optimum control moves (Understanding Model Predictive Control, n.d.).
1.6 Building Archetypes and Campus Building Modeling
Since the research objects of TrojanSense are students and faculty in USC campus, it is necessary to build the model
in university campus scale. Although the building modeling start from only one building: Watt Hall, the ultimate goal
for this project is extending the simulation scale from single building to the university campus. Thus, the approach for
district scale simulation modeling was explored.
Building Archetypes is a typical measured way used in energy modeling of a large group building. A group of
buildings with similar properties will be defined into same archetype group according to building shape, age, use,
climate, construction assemblies and building systems. (Reinhart and Davila, 2015).
1.7 Software
The software used for this research including rhino for creating model, grasshopper as a parametric modeling tool,
Meerkat GIS for generating university campus model, honeybee for supporting carbon-aware model, EnergyPlus and
openstudio as energy simulation tool.
1.7.1 Rhino and Grasshopper
Rhino is a 3D modeling software package that enables you to accurately model your designs ready for rendering,
animation, drafting, engineering, analysis, and manufacturing. Grasshopper is a visual programming language that
runs within the Rhinoceros 3D application. It is mainly used for parametric modeling in structural engineering and
architecture, lighting and building energy consumption. Unlike other programming software such as RhinoScript,
Grasshopper requires no knowledge of programming or scripting.
1.7.2 Grasshopper plug-in: Meerkat GIS and Honeybee
Meerkat GIS is a tool to generate building geometry in grasshopper from GIS database. GIS (Geographic Information
System) is a system combined with geography and cartography, using for geography data collection, storage, search,
analysis (esri, n.d.). By export the 2D map from GIS database and extrude them based on each building height, the
building geometry will be created for energy simulation.
Ladybug and Honeybee are two open source environmental plugins for grasshopper, which focus on environmental
architectural design and simulation. Ladybug mainly aims at lighting simulation by import standard EnergyPlus
weather files (epw.) into ladybug and then run various simulation including sun-path, wind-rose, radiation-roses,
radiation analysis, shadow studies and view analysis. Honeybee intend to make energy simulation available in a
parametric way by connecting grasshopper to EnergyPlus, Radiance, Daysim and OpenStudio for building energy
simulation.
1.7.3 EnergyPlus and Openstudio
EnergyPlus is a whole building energy simulation software for engineers, architects and researchers to model energy
consumption in all aspects: cooling, heating, ventilation, lighting, electricity, water use, etc. The program is based on
the user’s description of the building information specifically. Compared to OpenStudio, EnergyPlus is not a user
interface. It is an algorithm-based production which can be redirected into interface feature development.
OpenStudio is an interface for creating EnergyPlus input files, and it uses the .osm file extension. There arevarious
parameters to be determined in this interface including space type, thermal zones, HVAC system type and schedule,
construction detail and internal gains, etc. OpenStudio can achieve customized HVAC system for building energy
simulation while EnergyPlus focus on ideal air system. Both of them are important for this research.
1.8 Summary

7
TrojanSense is an user’s-participatory temperature adjustment tool for energy conservation, which serves USC
students and faculty. However, since users are unaware of the effects of temperature change, a carbon aware thermostat
is designed for making TrojanSense more comprehensive. This chapter explained several technical terms and their
significance to the research. A Thermostat is used to control the temperature of their heating/cooling facilities to match
the target value and a virtual thermostat is generated for occupants to report their willingness to adjust the room
temperature. The purpose of this research is to generate a carbon-aware virtual thermostat for TrojanSense and enable
occupants to understand carbon impacts. Since the research is based on a university campus scale, it is necessary to
research the approach for large numbers of buildings. Archetypes is a widely used method in urban scale energy
modeling, which can be applied in campus building energy modeling in the similar method. This chapter also
introduced some software, which includes Rhino, Grasshopper, Meerkat GIS, Honeybee, EnergyPlus and Openstudio.

8
2. BACKGROUND AND LITERATURE REVIEW
Thermal management approaches and temperature policy are introduced in this chapter to understand the significance
of having set point temperature control strategies in a university campus. There are several universities that conducted
similar studies in the participatory sensing field, however, most of them had a limited number of participants, buildings
or research period. TherMoostat, as a more comprehensive study of having occupant-participatory temperature
adjustment approach in University of California, Davis, is explained and discussed in this chapter. Compared with
TrojanSense project, the limitation of TherMoostat is that it has no solutions for spurious or exaggerated feedback. To
simulate university campus energy consumption, the study of urban scale building energy modeling is explored and
explained in this chapter. This approach can be applied in district scale building modeling including campus scale.
2.1 Thermal Management Approaches in Other Universities
As the largest sustainability conference in North America, the Association for the Advancement of Sustainability in
Higher Education’s (AASHE’s) annual conference aims at sharing effective information and actions to improve the
development of sustainability in higher education and surrounding communities. The thermal management approaches
shared by other institutions are valuable and useful for TrojanSense project.
2.1.1 Eastern Mennonite University: A Green Nerd in FM
Eastern Mennonite University is exploring programming changes to increase energy efficiency. For instance, they are
trying to create automated HVAC schedule based on University’s room scheduling system since the traditional HVAC
schedule is unreasonable which always depends on a fixed time. The other solution they came up with is changing
heating and cooling set points based on outdoor air temperature (AASHE Conference, 2018).
2.1.2 Wesleyan University: Creating a Successful Campus Temperature Policy
Sixty percent of energy and carbon emissions at Wesleyan University is accounted for by heating and cooling. The
Utilities Manager aims at reducing energy consumption by 20% in 2013 or 1.6 million dollars cost saving. In order to
address this goal, the Energy Conservation Policy was established by Wesleyan University to identify target
temperature ranges for campus buildings. This project has already had energy consumption reduction in four years.
Before this policy, Wesleyan University had spent millions on energy conservation over the past 15 years, but the
participation of students and faculties were limited until the Energy Conservation Policy was instituted. Currently,
each department, faculty, student is invited and plays an important role in the development of energy conservation
(AASHE Conference, 2018)
2.1.3 University of Maryland: Occupant Engagement and Centralized Controls
The University of Maryland designed a hybrid energy management solution to reduce the energy consumption caused
by devices plugged into wall outlets, called plug loads (PL). A novel PL management technology (Keewi Inc.) is able
to monitor energy consumption when plugged on devices. The occupants can view their personalized energy data
through the Keewi mobile app after the data collection period. Then the participants joined in the game of plug loads
energy management with points, rewards and team competitions to address energy reduction. The results indicated
that there is a 32% daily energy reduction during intervention compared to baseline for occupations using Keewi app.
27 staffers showed energy savings with Keewi management system our of 31 participants. Thus, the hybrid strategy
with socially-driven and centralized plug loads management is effective in higher educational facilities (AASHE
Conference, 2018)
2.1.4 Temperature Policy
As mentioned before, almost every institution of AASHE has its own campus temperature policy. The table 1 shows
the comparison of various universities temperature policy. For most buildings, the temperature policy is generated
after research according to the American Society of Heating, Refrigeration and Air conditioning Engineers (ASHRAE)
Standard 55(Erickson and Cerpa, 2012). The purpose of these policies is to clarify temperature expectations in indoor
spaces and prevent inefficient and wasteful operation of centralized equipment from occurring. In addition to energy

9
efficiency, university set point temperature can also avoid unnecessary utility costs associated with energy
consumption (Columbia University, 2006).
Besides temperature policy, the sustainability department of each university can issue guidance to help students make
a contribution to their own thermal comfort, such as wearing seasonally appropriate clothing, making sure windows
opened in appropriate ways for different seasons, making sure thermostats is not blocked by furniture or equipment,
etc.
University Heating Season (F) Cooling Season (F)
Harvard University 68-71 74-76
Columbia University 66-72 74-78
Bentley University 68.5-75 75-80
Tufts University 68 78
Rice University 68-72 74-78
University of Southern California 68 76
Table 1: Temperature Policy in Different Universities
2.2 Occupant-Participatory Temperature Adjustment Approach
The basic occupant-participatory temperature adjustment approach is asking for thermal votes via a mobile app or
web. Some similar researches have been discussed and most of them focus on university buildings. However, most of
them have limited research periods or participation. For example, Balajiy et al (2013) researched the 65 occupants’
thermal feedback of one university building in ten days. Erickson and Cerpa (2012) did the similar study with 39
participants in 7 thermal zones within a single building for five months. Both of these two studies proved that the
occupant-participatory temperature adjustment approach can be helpful for energy efficiency (Balajiy et al, 2013;
Erickson and Cerpa, 2012)
A more comprehensive study in this field is TherMOOstat, which aims at improving comfort and efficiency by
soliciting occupants’ thermal comfort votes in the University of California, Davis. Similar to TrojanSense project,
therMOOstat has two platforms for collecting users’ feedback: a widget on the university web portal and a mobile
application. For both app and widget interface, users need to submit their location (building and room number)
manually on the interface at first. For next step on widget, a question “how does the room feel” is raised, answers
include Hot, Warm, Perfect, Chilly, or Cold. Then users can leave any comments in an open-response format. For app
voting, after entering location, the overall satisfaction is questioned to all users. Response options are Very Dissatisfied,
Somewhat Dissatisfied, Neutral, Somewhat Satisfied and Very Satisfied (Figure 5-c). Then on the next screen,
occupants are assigned randomly to either thermal comfort vote or asked if they are willing to vote a change in campus
thermostat for energy conservation. If users vote yes, they are asked to select a temperature change of +/-1, 2 or 3
degrees (Figure 5-d). At the end of temperature voting, a summary of all conserve votes is provided to users
(Sanguinetti et al, 2016).

10

Figure 5: Thermoostat User Interface Screens (Sanguinetti et al, 2016)
In the TherMoostat project, the first stage of feedback collection from student, staff and faculty users lasted sixteen
months with 10,315 votes in total. The comfort and satisfaction results reflected that users more frequently reported
feeling “chilly” and “cold” rather than “warm” and “hot”, and most occupants voted “somewhat dissatisfied” or “very
dissatisfied”. For the willingness to conserve, the voting results indicated more than a third of users are satisfied with
the current temperature. The same percentage of users were willing to have 1 or 2 degrees temperature change (Figure
6). (Sanguinetti et al, 2016) Compared to the TrojanSense project, TherMoostat mainly focused on students’ thermal
sensation and satisfaction. However, TrojanSense is aiming to optimizing current HVAC set point temperature by
collecting occupants’ thermal feedback and raising occupants’ carbon awareness. Both of these two projects have two
platforms (web and mobile app) for occupant-participatory thermal sensing. Moreover, compared with other similar
studies, these two projects have longer research period, more research buildings and participants.

11

Figure 6: Willingness to conserve (Sanguinetti et al, 2016)
2.2.1 Exaggerated votes
So far, however, there has been little discussion about exaggerated votes and energy efficiency. Although some studies
mentioned the fact of exaggerated votes, research has not yet to systematically discuss the solution of how to solve
this problem. In TherMoostat, the researcher realized the users can vote exaggeratedly in order to get a stronger
response from HVAC system and they discussed this kind of behavior will lead to unnecessary energy consumption
(Sanguinetti et al, 2016). As figure 1d shows, a polar bear standing on a piece of ice and asking users “are you willing
to vote this”. This is an example of carbon aware education while in an obscure way.
Some other studies also suggested this problem, Erickson and Cerpa (2012) argued that a number of occupants may
provide fake feedback by over-inflating estimating their thermal comfort level. Users are too enthusiasm to providing
their feedback, which can be a double-edged sword. Some participants hope to vote more weight and thus receive a
more comfortable indoor environment. Another research proposed a strategy-proof thermal comfort voting model.
This framework is based on the idea that occupants can easily have motivations to provide spurious feedback (Zhang,
Lam, and Wang 2014).
2.3 Urban Scale Building Energy Modeling (UBEM)
Nowadays, the energy simulation is not only focused on one single building but in a large scale. In this project, carbon-
aware thermostat was used for collecting user’s feedback and telling user how much carbon emissions will cause in
specific thermal zone that the user is currently in. Since the energy results are based on zone level, thus it is necessary
to have large scale building energy model to determine user’s exact location and telling them correct carbon impact.
Urban scale building energy simulation is required for urban planners and policy makers to evaluate the impacts of
future scenarios, such as potential urban retrofits or new construction. One the other hand, the large datasets of metered
energy use, required for statistical models, are rarely available to modelers. One commonly accepted method to address
these difficulties is using ‘’bottom-up’’, refers to UBEM (Urban Building Energy Modeling) method. The general
workflow of UBEM including the following steps: (1) Collecting and analyzing of data input, (2) generating and
simulating of thermal models, and (3) result analysis and validation (Davila, 2017).
2.3.1 Collecting and analyzing of data input
For data input, all the collected information can be summarized into three main sets: Weather information, building
geometry and non-geometric model parameters (Davila, 2017). Weather data can be acquired from EnergyPlus web.
Building geometry can be generated by combining GIS (Geographic Information System) shape files with building
heights (Reinhart and Davila, 2015). As for non-geometric model parameters, including building usage, shape, floor
area, age, construction, HVAC system, and so on, are a large amount of building datasets need to be classified. As
mentioned before, building archetypes, as a most common strategy for UBEM, is widely used when lacking building
data. In building archetypes, buildings are classified into various groups in terms of their parameters. Then, a set of
complete templates including all thermal properties have to be defined to represent classified archetypes. The
templates can be either virtual or real sample.

12
2.3.2 Generating and simulating of thermal models
For the second step, the building is split into thermal zones for energy simulation. As Figure 7 shows, Davila (2017)
mentioned several approaches to split buildings. In the simplest case, each floor of a building can be considered as a
zone. Moreover, according to ASHRAE 90.1, the core and perimeter approach can be applied considering building
orientation. Shoeboxer zone abstraction method is another approach to maintain accuracy while reducing the amount
of simulation. In most cases, the thermal models are generated using GIS and BIM with custom scripts.

Figure 7: Thermal zoning approaches for 3D massing in UBEM (Davila, 2017)
2.4 Summary
In this chapter, thermal management approaches in three different universities were introduced. The set point
temperature policy in different universities were listed and compared with USC. The purpose for these background
researches is understanding the trend of sustainable development on the university campus. The research on university
temperature policy also provides a basic understanding of set point temperature range. In order to add occupants’
opinions to the set point control approach, some studies tested the feasibility of occupants’ thermal votes, which called
occupants-participatory temperature adjustment. Although most of the studies has limited research period and
participants, the results of these researches demonstrated participatory thermal sensing can improve energy efficiency
in university buildings. TherMOOstat, as an advanced project of thermal comfort feedback, was introduced and
explained in a detail way in this chapter. Some researchers have noticed untrue or exaggerated votes that may influence
the results of data collecting, however, no study discussed the reason and solutions for this problem systematically.
Thus, the idea of a carbon-aware thermostat in the TrojanSense project makes more sense to the field of participatory
thermal sensing because it refines the gap between thermal comfort and energy efficiency. This chapter also includes
research about urban scale building energy modeling approach, which can be used in university campus building
energy modeling to figure out where in the campus the occupants is and tell the user the correct carbon impacts of the
zone.

13
3. METHODOLOGY
3.1 Goal
The project aims at creating a real-time simulation-based feedback mechanism to tell users of a mobile app what the
carbon impacts will be from a given change (increase / decrease) in the current room temperature. When the user
adjusts the virtual thermostat by 1 degree, the user sees the annual increase or decrease in carbon emissions for that
specific zone that user is currently in. So that is why there needs to be a campus wide energy model. Because people
using the app are walking around the campus, but the feedback is on the zone level, so the app needs to know where
in the campus the user is, to determine the correct carbon impacts.
Figure 8 shows the basic work flow. In the whole process, users submit how much temperature set they were willing
to change from TrojanSense. This step is already been processed in the TrojanSense project, and the other steps are
the updates for TrojanSense to generating carbon-aware thermostat instead of only virtual thermostat. After collecting
users’ feedback of adjusting setpoints, the carbon-aware model was generated from the simulation results referring to
carbon emissions. The simulation results are monthly and annual cooling and heating load with different set point
temperature. After exporting these results into the TrojanSense mobile app via the table format, the carbon-aware
model is created. In this loop, the energy simulation is based on different users' feedback, and the simulation results
will influence the set points decision making. For instance, when users trying to vote “plus 5 degree” when three-
degree increase is enough for their thermal comfort, TrojanSense will remind him that how much carbon emission
will increase or decrease compared to the carbon emission based on current setpoint. Then the user might reconsider
the vote because the carbon emission education remind them that they were doing not environmentally.

Figure 8: Overall Methodology

The final goal is upgrading the user’s interface. As figure 9 shows, on the left is the existing mobile app interface of
TrojanSense. Users can adjust indoor set point temperature virtually by selecting how many degrees they prefer to
increase or decrease. On the left is the upgraded interface. Users can know the current set point temperature for both
heating and cooling and select their adjusted set point temperature by using the slider. Moreover, at the bottom of the
interface, the percentage of carbon increase or decrease will appear to remind users after adjusting setpoints. Therefore,

14
users can choose if they can sacrifice their thermal comfort to achieve carbon saving or not before submitting their
votes.

Figure 9: Before and After Upgrading User’s Interface

3.2 Approach
Figure 10 introduces the methodology in a more comprehensive way. There are three main parts of the project: data
collection; energy simulation and results visualization. These three parts are performed in order.

15

Figure 10: Detailed Methodology
3.2.1 Data Collection
Users’ feedback: As mentioned before, the data collection from users, refers to virtual thermostat adjustment or
occupants participatory setpoints adjustment, as the core objective of the current TrojanSense project, was started in
2018 and already collected some feedbacks from users. To achieve data collection from users, firstly, Bluetooth
sensors called Beacon were installed in different rooms on the USC campus including a lecture hall, seminar classroom,
study room, library, etc. Since the signal range of the Beacon is limited to five meters, researchers need to make sure
each room has at least one beacon for feedback collection. In order to make sure as many as possible users can detect
the signal automatically without losing any, the best way is hiding beacon in the center of a room. Next, when users
turn on their mobile Bluetooth, the TrojanSense mobile app could detect their room location automatically, or they
can manually enter which building and room number they are currently in. Then users can adjust the temperature via
website or mobile application. During data collection period, the building model information, the outside climate data
should be collected as early as possible. The users’ set points feedback, since it is a long-time work, can be processed
in parallel to the energy simulation.
Building Information: Before starting an energy simulation, it is necessary to have enough building information to
generate the building energy model. Because this is a study on the USC campus, most building information can be
obtained from USC Facility Management Services (FMS). The required building information for 3D modeling
includes building location, floor numbers, floor area, window to wall ratio, story height. Outdoor climate data should
be prepared for energy simulation and can be downloaded from the EnergyPlus official website.
3.2.2 Energy Simulation
After adequate preparation work, the collected data can be input into an energy simulation model. The building model
should be redrawn in Rhino since the model from FMS is too complicate to run energy simulation in Grasshopper.
The detailed information such as staircase and furniture should be deleted to get a clear and simple model for energy
simulation. The revised building model has the correct location and building geometry with windows and accurate
floor height. The climate data should be input into honeybee in grasshopper. The thermal zones for each floor can be
defined in Honeybee by using component “split thermal zones into perimeter and core”. Another important
information is the type of HVAC system, the type and schedule are two main parameters to be determined when
inputting into energy simulation tools. The energy simulation will be processed in EnergyPlus and Openstudio, which

16
will be introduced later in section 3.4.1. The simulation results are the cooling load, heating load, electricity and
lighting; the study will focus on the energy consumption and carbon emission caused by analyzing heating and cooling
load.
3.2.3 Results Processing and Visualization
The data processing and results visualization are important parts for carbon-aware education. The causing or saving
of carbon emission should be visualized in a simple and direct way. Too complicated data may confuse occupants’
understanding the meaning of set point temperature adjustment while too simple data may not raise their carbon saving
awareness.
For data processing, the energy simulation results should be converted to carbon emission ultimately since both energy
consumption and carbon emission results are important for a carbon-aware thermostat model. The data need to be
translated into table format in Excel for using in the TrojanSense app. For data processing of each zone, it is a monthly
energy consumption and carbon emission table with 12 rows and 66 columns. Each row represents a specific month
in a year while each column is a setpoint combination (heating and cooling setpoints). Each cell indicates the energy
consumption or carbon emission results under then given set point temperature in a specific month. (see more details
in section 3.4.4)
When the ultima real-time simulation carbon aware model was input into the TrojanSense mobile app, users are able
to be aware of how much carbon impact they are causing or reducing by changing set point temperature. Users can
understand how much energy consumption and carbon emission will be generated orsaved in the current hour when
they vote a set point temperature in the zone where they are in. Occupants can also select a set point temperature and
a specific time of a year to see the environmental impacts. Moreover, the results can also tell the app user for a given
zone and an initial set point temperature, what the carbon impact would be over the next period (e.g. next month, next
6 months, next year) if any arbitrary (e.g. increase 2 degree) change in the set point temperature.
For website visualization, uses can choose the building from the campus map in which they would like to explore
more energy information. Moreover, the carbon emission of different thermal zones can be viewed. When occupants
get more knowledge about energy consumption, their environmental awareness will be raised. Thus, occupants can
consider both thermal comfort and energy efficiency when adjusting set point temperatures when voting for their
comfortable temperatures.
3.3 Static Energy Model (Hand Calculation)
Before running energy simulation, a static energy model calculating room cooling load was generated based on
formulas and standards. The following information shows the calculation formula. (Figure 11)
3.3.1 Heat Gain by Natural Ventilation
Heat Gain by Natural Ventilation = SH x Density x VHC x ∆T x CFM
SH = Specific Heat; VHC = Volume Heat Capacity; ∆T = Indoor and Outdoor Temperature Difference;
CFM = Ventilation Rate (Cubic Feet per Minute) (Stein, 2006)
3.3.2 Internal Heat Gain
Internal Heat Gain (People) = Number of People x Sensible Gain per Occupants
Based on ASHARE 90.1, the sensible heat gain of light office work is 400 Btu/h per occupants.
Internal Heat Gain (Lighting) = watt x 3.41 Btu/h/watt x ballast factor
Internal Heat Gain (Equipments) = Number of Occupants x Heat Gain from Laptop (Stein, 2006)
Assume 10 Btu/h per laptop.
3.3.3 Heat Gain through Envelope (Roof, Walls and Floor)

17
Heat Gain (Envelope) = U x A x DETD (Stein, 2006)
U = U-Vale; A = Surface Area; DETD = Design Equivalent Temperature Difference
The ultimate calculation result is the sum of various heat gains above in Btu/h. In order to get carbon emissions caused
by the cooling load, the energy consumption needs to be converted to carbon footprints.

Figure 11: Static Model (Hand Calculation)
3.3.4 Grasshopper
As mentioned before, Grasshopper, as a graphical algorithm modeling tool, is tightly integrated with Rhino, which is
a 3D modeling tool widely used in the architecture design field. In Grasshopper, programs are generated by using a
node-based interface. Components are dragged out of palettes and placed onto the canvas. Each component node
represents a certain function with input and output data. Correlative components can be correlative with each other,
which makes grasshopper become a better choice for real-time parametric energy modeling for this research. There
are various existing component palettes in grasshopper, however, for this project, additional plug-in called Meerkat
GIS and Honeybee need to be installed for energy simulation. The Meerkat GIS can be used to generate whole campus
building model by extrude 2D plan in terms of their heights. The weather data, building schedule, HVAC system and
other building information related to energy simulation were all set up in Honeybee.
All the static energy model mentioned above and following dynamic energy model were generated in grasshopper.
For the static energy model, the calculations result of formula is based on number of occupants in the zone. For
instance, to calculate the heat gain generated by natural ventilation, the formula “Heat Gain by Natural Ventilation =
SH x Density x VHC x ∆T x CFM” was used (see more detail in 3.3.1), in this case, the density is the total number of
occupants divided by total floor area. As the variable, the number of occupants influence the density and eventually
affect the results of cooling load. When the number of occupants increase, the heat gain caused by natural ventilation
rise, the total cooling load also increase. This is the parametric model for hand calculation of energy consumption.
However, there is an important disadvantage of this model, which is lack of time and set point temperature, so results
cannot change dynamically at different times in a year or under different set points of HVAC system. To bridge this
gap, a more dynamic energy model is more suitable for carbon-aware thermostat.
3.4 Dynamic Energy Model

18
In the simple static model, the interior heat gain is based on floor area, which means that when the floor area changes,
the number of occupants and ventilation rate will be influenced as well (plug load and lighting). This is the general
concept of parametric model. However, the limitation of the static state model is it only calculate energy consumption
in a specific hour. It is too complicated to calculate the annual energy consumption using static state model. Compared
to the static model, the dynamic energy model can be more parametric, which is a benefit for real-time energy
simulation. It means even if the set point temperature is the constant, users can still get different energy consumption
and carbon emission at different times of year. On the other hand, when the set point temperature changes, even in the
same time period, the energy consumption and carbon emission will correspondingly change. For example, in the
cooling season, when the set point temperature increases, the energy consumption and carbon emission will decrease,
which is a benefit for sustainable development. In the heating season, it has the opposite situation.
To generate a dynamic energy model, rhino, honeybee in grasshopper and energyplus are the three main software
programs. In this process, the general building model is created in rhino. Weather data and building information
including window to wall ratio, construction detail, HVAC system and schedule, set point temperature are input in
Honeybee. EnergyPlus is used for energy simulation based on different times.
3.4.1 Building Energy Modeling in Honeybee and EnergyPlus
As mentioned before, Ladybug and Honeybee are two open source environmental plugins for grasshopper, which
focus on environmental architectural design and simulation. Ladybug mainly aims at lighting simulation while
honeybee focus on making energy simulation available. For this study, Honeybee is more important and widely used
since the role of energy simulation. Some component for weather data and results visualization in Ladybug were also
used in this research.
Figure 12 is the detail workflow for building energy modeling with the icon and name for each step. There are 11
steps from generating building model in Rhino to run energy simulation using Openstudio or Energyplus as the
platform. The first two steps focusing on creating a simplified model for energy simulation and split this model into
different thermal zones. From step 3 to step 8, the parameters, including window to wall ratio, construction detail, set
point temperature, HVAC system type were applied to the model. Weather data was assigned to each zone after setting
up all the building information in step 9. Step 10 defined the simulation output and timestep. Energy simulation can
be done after all these steps. The following sections will discuss details about each step.

Figure 12: Workflow for Building Energy Modeling
Before running energy simulation, in addition to download weather files (epw.) from EnergyPlus official website, the
building construction detail, window to wall ratio, HVAC system and schedule of Watt Hall were acquired from USC
facility management service. As mentioned before, after drawing the basic building model in rhino, first thing needs

19
to do is using component “Brep” to generate it in honeybee. Secondly, the window to wall ratio can be defined for
each façade. Then the building model should be split into thermal zones using perimeter and core approach. For
example, in Watt Hall, there are four floors including basement with five thermal zones (south, north, west, east, core)
on each floor, so the total number of thermal zones in Watt Hall are 20 zones. The energy simulation results are
different for each thermal zone even if the set point temperature and time are same. Then the building schedule should
be defined for these thermal zones, for Watt Hall, all the zone can be applied with classroom schedule in Honeybee.
Next, the type of HVAC system and system setpoints should be defined for these zones. By using the slider component
in grasshopper, the set point temperature can be adjusted into different value (Figure 13). The range of the set point
temperature is determined by occupants’ feedback from TrojanSense app or web. The basic method of simulating
energy consumption is connecting required building information and weather data in Honeybee with EnergyPlus as
simulation tool. The simulation results can be presented in hourly, daily, monthly or annually in grasshopper. The
simulation results include cooling load and heating load are required for this research.

Figure 13: Dynamic Energy Model of Watt Hall
3.4.2 Building Energy Modeling: Watt Hall Model
The first two steps for building energy modeling in this study is generating simplified building model and splitting it
into thermal zones. As mentioned before, Watt Hall, as the headquarter of school of architecture in USC, was selected
as the exemplary building for energy modeling. The model of Watt Hall was simplified in order to generate appropriate
model for energy simulation in Honeybee and EnergyPlus (Figure 14). Figure 14-a is the original model of Watt Hall
in Rhino. This model was created by USC Facility Management Serves (FMS) and included all the detailed
information of Watt Hall. Table 2 is the building area and floor height information obtained from the original model.
However, for building energy simulation, the requirement for modeling is different from architectural modeling.
Honeybee is unable to define the envelope with thickness. Some other building information, such as staircase, lighting
fixtures, and furniture are also not important and may bring inconvenience for energy modeling. Thus, a new model
was created in Rhino based on the building information from original model (Figure 14-b). The unnecessary
information was ignored in this model. The window size was determined by window to wall ratio, which is more
convenient and faster for running energy simulation. However, this model is not suitable for the study because of the
convex geometry in the third floor and “split into thermal zones by perimeter and core” in Honeybee only can
recognize non-convex geometry. Although this convex geometry does not influence the simulation results on building-
level, it is unable to generate zone-level simulation results, which is important for this study because the occupants’

20
temperature adjustment is based on room level, so the simulation results of a thermal zone is more appropriate for
occupants to understand how much energy will increase or decrease in this space. To satisfy this requirement, a more
general model without convex geometry is created (Figure 14-c). Figure 14-d is the model after setting up window
size and splitting into 20 thermal zones in terms of core and perimeter approach. Figure 15 shows the process of “Brep”
building model from Rhino to grasshopper, which is the first step of creating building energy model in grasshopper.
Then the model was split into 20 zones with 5 zones for each floor.
Basement First Floor Second Floor Third Floor
Area (sqft) 23449.5 18407.25 18407.25 24017.5
Height (ft) 19 14.5 14.5 22

Table 2: Building Floor Area and Height

Figure 14: Process of Building Energy Modeling – Watt Hall

Figure 15: “Brep” Model from Rhino to Grasshopper and Split Thermal Zones
3.4.3 Building Energy Modeling: Setting up Parameters

21
Before running energy simulation, it is important to customize template for the target building. Although there is
default template for school energy simulation in Honeybee, as an architectural school, the schedule of Watt Hall is
different from other schools because studios are opened for students all through the day. One the other hand, since
Watt Hall was built several decades ago, the building energy standard was much more different at that time than
nowadays, which leads to large discrepancy on simulation results if using current default template. Generally, the
customized template for building energy modeling including window to wall ratio, construction detail, occupancy
schedule and load, HVAC system and set point temperatures. The following figures show general building information
of Watt Hall and how the information was defined in grasshopper and Honeybee.

North Facade South Facade West Facade East Façade
Window to Wall Ratio 0.6 0.6 0.5 0.5
Table 3: Window to Wall Ratio

Figure 16: Set Up Window to Wall Ration in Honeybee
For the building construction detail, redefined materials of wall, floor, roof, and window enable the more accurate
simulation output. Since the Wall Hall was built a few decades ago, the thermal resistance for the building envelope
are not so well compared with nowadays building materials. For instance, as table 4 shows, the wall material of Watt
Hall includes 1’’ stucco, 8’’ concrete and ½’’ gypsum with no insulation layer. Figure 17 shows the components for
customize building materials and how to connect them. To redefine a building material, which energy modeling
standard are used and which climate zone the building is in should be determined firstly, then use “Set EP Construction
Zone” component to modify the construction details (Figure 17). For Watt Hall, the building is in the ASHARE climate
zone 3C and the layers of building construction material are defined in terms of CBESC 1980-2004 building energy
modeling standard.

Wall Floor Roof Window
Layers 1’’ Stucco
8’’ Concrete
½’’ Gypsum
½’’ Gypsum
Insulation-1.76
½’’ Gypsum
Roof Membrane
Roof Insulation-2
Metal Decking
Clear 3mm
Air 13mm
Clear 3mm

22
U-Value 0.64999 0.116004 0.097332 0.417202
R-Vaule 1.538484 8.620409 10.274124 2.39692
Table 4: Construction Detail

Figure 17: The Process of Customizing Building Construction in Honeybee
As mentioned before, although there is occupancy schedule template for schools in Honeybee, the schedule
indicated that the building was not occupied in the evening and during the weekend, which is different from a
schedule of an architecture school, so the building occupancy schedule is modified as well. Table 5 is the
customized building occupancy schedule for Watt Hall. Figure 16 shows how to generate a new occupancy
schedule: firstly, create a one-day schedule for Monday to Friday and multiply it by 5. Then generate another one-
day schedule for weekend and multiply it by 2. Next combine them to get a weekly schedule and multiply it by 52
(Figure 18).
Monday to Friday Weekend
Occupancy Density
8:00am-6:00pm: 0.7 8:00am-6:00pm: 0.2
6:00pm-2:00am: 0.2 6:00pm-2:00am: 0
2:00am-8:00am: 0 2:00am-8:00am: 0
Table 5: Building Occupancy Schedule

Figure 18: Customize Building Occupancy Schedule

23
After modifying occupancy schedule and load, the HVAC system and its set point temperature should be defined for
the building. The component named “Assign HVAC System” can choose and apply different HVAC system to the
building (Figure 19).

Figure 19: Choose HVAC System
Each energy simulation result is based on a group of cooling and heating set point temperatures. The range is from 70
to 78 degrees for cooling set point and heating set point temperature vary from 66 to 74 degrees with 1-degree intervals
(Table 6). Each group consists one heating set point and one correspond cooling set point. The rule is the heating set
point should be lower than cooling set point. In Table 7, the check mark means this combination is feasible for energy
simulation. Table 7 indicates all the possible ways for combination. There are 66 combinations for cooling and heating
set point temperature. Figure 20 shows the process of setting up cooling and heating set points. The temperature units
used in Honeybee is degree Celsius, so the formula C=(F-32) *5/9 is applied to convert degree Fahrenheit to degree
Celsius.

Cooling Set Point Heating Set Point
Temperature (℉) 70-78 66-74
Table 6: Heating and Cooling Set Point Range

Cooling
Heating
70 71 72 73 74 75 76 77 78
66 P P P P P P P P P
67 P P P P P P P P P
68 P P P P P P P P P
69 P P P P P P P P P

24
70 O P P P P P P P P
71 O O P P P P P P P
72 O O O P P P P P P
73 O O O O P P P P P
74 O O O O O P P P P
Table 7: Cooling and Heating Set Point Combination

Figure 20: Adjusting Heating and Cooling Setpoints
The next step is adding weather data for energy simulation. As mentioned before, the weather data can be downloaded
from EnergyPlus website.

Figure 21: Adding Weather Data
3.4.4 EnergyPlus to TrojanSense Data Transfer Format
The simulated results in EnergyPlus cannot be input into TrojanSense APP directly, however, a table format in Excel
is an appropriate way to bridge this gap. Firstly, the simulation results should be recorded by automating different set
point temperature combinations. This is achieved here by using the component “fly” in grasshopper. Then the
simulation data can be exported from grasshopper into csv. files. Each row in a file represent the time, it can be an
hour, a day or a month in a year. For annual energy consumption, it is not convictive to only include cooling load
under different cooling set points. The heating load should be considered as well because the results is annual energy
consumption. Thus, the simulation output is cooling and heating load at different combinations of cooling and heating

25
set points. For each thermal zone, there should be two csv files (cooling load and heating load csv files) for every
simulation with different set point temperature. In the Figure 22, every two columns represent a group of heating and
cooling set points. Each row is the cooling/heating/cooling and heating load at different months. Each cell represents
the heating or cooling load under a specific combination of set points. For instance, when the cooling set point is 70
degree and heating is 66 degree, the number in the blue cell means there are 12350.12KBtu cooling load in January
and red cell is the heating load. Green cell indicates the total cooling and heating load in January is 14108KBtu under
this set point combination.

Figure 22: Data Processing in Excel – Monthly Cooling and Heating Load for Zone 16
3.4.5 Background Processing for Carbon-Aware Thermostat
Zone-level energy simulation is the kernel for carbon-aware thermostat because it can correspond to different rooms.
Although room-level energy simulation results can be achieved using Honeybee and EnergyPlus, it is too complicated
for this research to gather each room data for the building and extend to entire campus buildings. Compared to room-
level energy simulation, generating zone-level energy simulation results is more convenient since thermal zones can
be split directly in Honeybee as long as the building model has no convex geometry. When users adjust set point
temperature on mobile APP or web, the virtual thermostat will recognize which thermal zone they are currently staying
and tell users the discrepancy of energy consumption and carbon emission by this temperature adjustment. Thus, the
first step for zone-level energy simulation is to figure out the room belongs to which thermal zone. Figure 23
(TrojanSense, 2019) are the rooms with beacon (signal detector) in Watt Hall and table 8 correspond each room to the
thermal zone. There is no beacon installed in any room on the first floor at this point because the first floor is lobby
and exhibition space with few people staying.

26

Figure 23: Rooms with Installed Beacon in Watt Hall (TrojanSense, 2019)
Basement
Room B1 ARCH Library B1 Studio
Zone Number 4 3 1

Second Floor
Room
2
nd
Floor
Office
Lower
Rosendin
Studio 200 Studio 208 Studio 209 Room212

Zone Number 12 10 13 11 14 10

Third Floor
Room
Upper
Rosendin
Room 324
3
rd
Floor
Balcony
MBS Corner
ARCH
Corner
Zone Number 15 19 16 16 18
Table 8: Corresponded rooms and thermal zones in Watt Hall
Figure 24 illustrated the relations between occupants’ feedback and background processing. On the left is the virtual
users’ mobile interface for carbon-aware thermostat, which already detected the user’s current location automatically.
The set point temperature is determined, and the user can choose the energy caused by which heating and cooling set
points combination they wish to know. If the user chooses higher heating set point than cooling set point, there will
be a warning to remind them that heating set points should be lower than cooling. The right side of the figure shows

27
how the location and set point information can be processed to generate the simulation output for uses. The room
which the user is currently in can be correspond with the thermal zone which the room is in, for instance, the Watt
Hall MBS Corner and the 3
rd
Floor Balcony are in the south zone of third floor, which is the zone no.16. Then in the
background processing, the simulation output of the No.16 thermal zone will be extracted from the database, which
including all the simulation results for each zone. As mentioned before, for the simulation output of each zone, it
includes 66 combinations of heating and cooling set point temperature. In this case, the simulation results under 71℉
cooling set point and 69℉ heating set point is needed for calculating how much percentage of carbon emission will
increase or decrease.

Figure 24: User’s Temperature Vote and Background Processing
Since the purpose is telling user how much energy they will be saving or generating with each adjusted set point
temperature compared with current set points and raising occupants’ carbon awareness, the final output should be the
percentage of how much energy consumption and carbon emission will increase or decrease when changing set points.
Therefore, when the simulation results are prepared, the current set point temperature is of the essence. According to
the USC room temperature policy, most rooms in USC campus are heated to 68℉ in the heating season and cooled to
76℉ in the cooling season (USCFMS, n.d.). On the other hand, the TrojanSense project has its own room temperature
monitoring approach, which is using Internet of Things (loT) sensors to detect real-time indoor temperature and
humidity. However, according to the TrojanSense temperature monitoring in Watt Hall, not every room has the
average temperature with 76℉ while some room temperature is in between 70℉ to 74℉ in cooling season, that is the
reason to use 72℉ as cooling setpoint instead of 76℉ (Dong, 2018). Figure 25 shows how to generate the carbon
emission increase/decrease percentage table. On the left is the energy consumption with current room set point
temperature (72℉ for cooling and 68℉ for heating) and the other is the new energy consumption with adjusted set

28
point temperature (e.g. 71℉ for cooling and 69 for heating). Both results can be extracted from the excel table. The
current annual energy consumption is constant in any cases while the new annual energy consumption is changed
based on users’ temperature adjustment. The formula in Figure 25 pointed out how to calculate the annual energy
saving or wasting percentage. If the calculation result is negative, which means more energy/carbon emission will
cause with the adjusted set points. In the opposite, positive result means the carbon emission will decrease. For
instance, in this case, there will be approximately 13.41% carbon emission caused compared to current situation if
adjusting the cooling set point to 71℉ and 69℉ for heating.

Figure 25: Process of Generating Carbon Increase/Decrease Percentage Table
3.5 Campus Scale Building Energy Simulation
Similar to urban scale building energy modeling, campus scale building energy simulation in this research consists of
two parts: thermal modeling and building data input. As figure 26 shows, to generate building geometry in Rhino, the
2D building footprint could be exported from Geographic Information System (GIS) data, then the building footprint
can be extruded from Meerkat GIS. According to core and perimeter method, the extruded building can be subdivided
into five different thermal zones, including south, west, north, east and core zones.
In addition to building geometry, building types is another important component in campus energy modeling. There
are various ways to divided buildings into different Archtype groups, such as construction, building system, geometry,
age, usage. The research select usage as the classification basis, dividing USC campus buildings into five groups,
including administration, library, dormitory, classroom and laboratory.

29

Figure 26: Workflow of Campus Scale Building Energy Simulation
3.6 Summary
The final goal is raising occupants’ carbon awareness by upgrading current users’ interface of temperature adjustment.
This chapter explained the methodology of generating real-time carbon-aware thermostat for the TrojanSense project.
The three main steps are 1) data collection; 2) energy simulation 3). Results visualization. The data collection from
users by mobile app or web via Bluetooth node were established at the beginning of the project. For the energy
simulation, the building information were collected from FMS and exported into rhino and grasshopper for building
energy modeling. The basic work flow of the Watt Hall energy modeling was introduced in this chapter. For the results
visualization, the process of transferring simulation output from grasshopper to TrojanSense using table format (csv.
file) and how does it present in mobile app were explained. This chapter also discussed the approach for campus scale
building energy simulation, which includes subdividing building into thermal zones and Archtype groups.

30
4. RESULTS
This chapter introduces the simulation results of the USC Watt Hall building for the purpose of illustrating the
applicability of the workflow for generating zone-level data for one campus building. The results are generated
according to the methodology explained in the previous chapter. The results include annual and monthly simulation
output for two typical thermal zones, each floor and the entire building with 66 various combinations for set point
temperatures. For reference, with the zoning scheme used by the workflow, Watt Hall is composed of 20 total zones.
Therefore, these two zones represent only a limited set of outcomes, but again are considered sufficient for
demonstrating the results that can be readily obtained using the modelling and simulation approach developed in this
thesis. The results are presented in both table and graph format to evaluate the effect of adjusting set point temperatures.
The table format is used to export into the TrojanSense mobile app and web while the graphs are illustrated as a more
effective for results visualization. The weather data of Los Angeles was also illustrated in this chapter, for better
analyzing and understanding simulation results.
4.1 Weather Data and Room Temperature
As an important factor for building energy simulation, weather data influences the results directly. The weather data
for this research consists temperature, humidity, wind and solar radiation data for all 8760 hours of a year, developed
by the US Department of Energy and obtained from EnergyPlus website as epw file (EnergyPlus, n.d.). Figure 27
illustrate the hourly dry bulb temperature in Los Angeles, which is measured when thermostat is exposed to the air
while not affected by radiation and moisture. The range of temperature in Los Angeles varies from about 40 degree to
almost 100 degree Fahrenheit in a year. According to the figure 27, more than half of the hourly temperatures in Los
Angeles are higher than 68 degree. From July to September, most of hourly temperatures are higher than 75 degrees.
The purpose of showing this data is to demonstrate that most of the time in a year in Los Angeles is cooling dominant,
which means the cooling season is longer than heating season. These weather data files represent typical
meteorological years, and not the increasingly warm weather conditions Los Angeles will likely expect to see in future
years. So, cooling energy savings from and extended setpoint range are likely to be underestimated.
Figure 27: Hourly Dry Bulb Temperature of Los Angeles
The figure 28 (Dong, 2019) shows the range of temperature for most rooms in Watt Hall, which indicates that the real
room temperature is not always the same as the setpoint temperatures. According to the USC room temperature policy,
the cooling setpoint is 76℉, however, most rooms have actual temperatures which are lower than 76℉., Some average
room temperatures are around 72℉ and others are close to 74℉. So it is necessary to have real-time room-level thermal
data from loT sensors to point out the discrepancy between real room temperature and campus policy.

31

Figure 28: Room Temperature Comparison (Dong, 2019)
4.2 Building-Level Energy Simulation Results
Building-level energy simulation is applied as the most basic method to investigate the overall energy consumption
for this building at different combinations of set points. The result is obtained by considering Watt Hall as a single
thermal zone during simulating process (i.e. without subdividing into multiple thermal zones). Table 9 and Figure 28
indicated the building energy consumption value under different combinations of cooling and heating set point
temperatures. The set point temperature combination resulting in the lowest and highest energy consumption can be
indicated as the reference results for adjusting set point temperatures. According to Figure 28, the X-coordinate
includes each group of set point temperatures. For each setpoint combination, the number in the front represents
cooling set point and the second number is the heating set point, for instance, the first setpoint combination is “70&66”,
which means the cooling setpoint is 70℉ while heating setpoint is 66℉. The Y-coordinate is the total annually heating
and cooling load (KBtu). The table 9 indicates the exact energy consumption with each setpoint combination. For each
“setpoint” row in the table, the cooling set point is constant while the heating set point is increasing with 1-degree
interval. Each column in the table has the constant heating setpoint with increased cooling setpoint. The highest and
lowest energy consumption with their setpoint combinations are highlighted in the table.
Setpoints 70&66 70&67 70&68 70&69

Cooling Setpoint & Heating Setpoint
Energy 2682108 2719456 2764015 2814872

Setpoints 71&66 71&67 71&68 71&69 71&70

Energy 2369966 2404966 2448910 2500807 2557092

Setpoints 72&66 72&67 72&68 72&69 72&70 72&71

Energy 2042795 2075908 2117059 2168296 2227806 2285373

32
Setpoints 73&66 73&67 73&68 73&69 73&70 73&71 73&72

Energy 1786193 1818197 1857021 1905595 1963828 2031352 2099514

Setpoints 74&66 74&67 74&68 74&69 74&70 74&71 74&72 74&73

Energy 1539103 1570210 1607800 1653683 1709188 1775484 1850915 1931658

Setpoints 75&66 75&67 75&68 75&69 75&70 75&71 75&72 75&73 75&74
Energy 1331326 1361653 1398217 1442044 1494746 1558058 1632561 1716669 1812632
Setpoints 76&66 76&67 76&68 76&69 76&70 76&71 76&72 76&73 76&74
Energy 1121972 1151393 1187231 1229725 1280194 1340467 1411653 1494794 1588522
Setpoints 77&66 77&67 77&68 77&69 77&70 77&71 77&72 77&73 77&74
Energy 956208.9 984925.8 1019732 1061111 1110263 1167958 1235964 1315317 1407942
Setpoints 78&66 78&67 78&68 78&69 78&70 78&71 78&72 78&73 78&74
Energy 807049.6 835047.1 869071.4 909438 957355.8 1013323 1078579 1155073 1243434
Table 9: Total Annually Energy Consumption of Watt Hall

Figure 28: Total Annual Heating and Cooling Load of Watt Hall at Different Set Points
The following points are key findings from Table 9 and Figure 28:
- When the cooling set point temperature is 70℉ while heating set point temperature is 69℉, the total
heating and cooling load has the greatest value compared to other 65 setpoints combinations.
- When the cooling set point temperature is 78℉ while heating set point temperature is 66℉, the total
heating and cooling load has the lowest value compared to other 65 setpoints combinations.
- When cooling set point temperature is constant and heating set point temperature gradually increase,
the total heating and cooling load increase.
- When heating set point temperature is constant and cooling set point temperature gradually increase,
the total heating and cooling load decrease.

33
4.3 Floor-Level Energy Simulation Results
The floor-level energy simulation aims to provide more detailed information about energy consumption for each floor
of Watt Hall in terms of different setpoint combinations. As mentioned before, there are four floors including one
basement in Watt Hall. The results are obtained by splitting the thermal zoning of the building into floors. Because
the area for each floor is not the same, the cooling and heating Energy Use Intensity (EUI) is used to compare the
energy consumption of each floor. The EUI (kBtu per sqft) is calculated by using energy consumption of this floor
(kBtu) divided by floor area (sqft). In each setpoint combination, the number in the front is the cooling setpoint while
the number behind is the heating set point temperature. For each table in this section, each “setpoint” row in the table
has the same cooling load with increased heating set point temperature, and for each column, every heating setpoint
is same while cooling setpoint is increasing with 1-degree interval. The highlighted cells are the highest and lowest
EUI with their setpoint combinations. For each figure in this section, the X-coordinate is the different combinations
of heating and cooling setpoints. The Y-coordinate is the annual heating and cooling EUI with same range (0-50 KBtu
per sqft) for each graph (Figure 29-32).

34
Setpoints 70&66 70&67 70&68 70&69

Cooling Setpoint & Heating Setpoint
EUI 23.49665 24.08046 24.79625 25.61623

Setpoints 71&66 71&67 71&68 71&69 71&70

EUI 20.50189 21.05156 21.75449 22.60924 23.53129

Setpoints 72&66 72&67 72&68 72&69 72&70 72&71

EUI 17.43844 17.95912 18.62945 19.46604 20.45122 21.44636

Setpoints 73&66 73&67 73&68 73&69 73&70 73&71 73&72

EUI 15.07146 15.57302 16.21045 17.01866 17.97805 19.10961 20.27825

Setpoints 74&66 74&67 74&68 74&69 74&70 74&71 74&72 74&73

EUI 12.84369 13.32771 13.94499 14.7138 15.65015 16.75902 18.03911 19.43053

Setpoints 75&66 75&67 75&68 75&69 75&70 75&71 75&72 75&73 75&74
EUI 10.91957 11.39164 11.98832 12.72396 13.62254 14.69214 15.95967 17.40246 19.08019
Setpoints 76&66 76&67 76&68 76&69 76&70 76&71 76&72 76&73 76&74
EUI 9.028543 9.479903 10.06472 10.77763 11.64205 12.67334 13.895 15.31759 16.96271
Setpoints 77&66 77&67 77&68 77&69 77&70 77&71 77&72 77&73 77&74
EUI 7.502325 7.942156 8.502315 9.195407 10.0412 11.0324 12.20755 13.57519 15.18802
Setpoints 78&66 78&67 78&68 78&69 78&70 78&71 78&72 78&73 78&74
EUI 6.136149 6.560845 7.106641 7.778815 8.599986 9.565726 10.69686 12.02976 13.57841
Table 10: Cooling and Heating Load of Basement

Figure 29: Cooling and Heating EUI of Basement

35

Setpoints 70&66 70&67 70&68 70&69

Cooling Setpoint & Heating Setpoint
EUI 31.972 32.31374 32.74545 33.26144

Setpoints 71&66 71&67 71&68 71&69 71&70

EUI 28.11183 28.43083 28.84016 29.34916 29.93534

Setpoints 72&66 72&67 72&68 72&69 72&70 72&71

EUI 24.04932 24.35061 24.72958 25.21432 25.81138 26.41065

Setpoints 73&66 73&67 73&68 73&69 73&70 73&71 73&72

EUI 20.87995 21.1698 21.52456 21.97625 22.54239 23.2309 23.93135

Setpoints 74&66 74&67 74&68 74&69 74&70 74&71 74&72 74&73

EUI 17.81653 18.09612 18.43623 18.86342 19.3878 20.04261 20.81586 21.62785

Setpoints 75&66 75&67 75&68 75&69 75&70 75&71 75&72 75&73 75&74
EUI 15.25052 15.52091 15.84876 16.25429 16.75358 17.36638 18.10445 18.96451 19.92529
Setpoints 76&66 76&67 76&68 76&69 76&70 76&71 76&72 76&73 76&74
EUI 12.65444 12.91549 13.23339 13.62331 14.09828 14.68047 15.37754 16.2033 17.15131
Setpoints 77&66 77&67 77&68 77&69 77&70 77&71 77&72 77&73 77&74
EUI 10.63491 10.88722 11.19566 11.57116 12.02992 12.58597 13.24782 14.03188 14.95385
Setpoints 78&66 78&67 78&68 78&69 78&70 78&71 78&72 78&73 78&74
EUI 8.823848 9.067968 9.367782 9.731988 10.17579 10.7112 11.34869 12.09347 12.9674
Table 11: Cooling and Heating EUI of the First Floor

36

Figure 30: Cooling and Heating EUI of the First Floor

Setpoints 70&66 70&67 70&68 70&69

Cooling Setpoint & Heating Setpoint
EUI 33.54905 33.89333 34.32247 34.82947

Setpoints 71&66 71&67 71&68 71&69 71&70

EUI 29.53838 29.86082 30.26963 30.7727 31.35185

Setpoints 72&66 72&67 72&68 72&69 72&70 72&71

EUI 25.3211 25.62649 26.00595 26.48832 27.07852 27.66666

Setpoints 73&66 73&67 73&68 73&69 73&70 73&71 73&72

EUI 22.02554 22.3199 22.67546 23.12476 23.6873 24.36328 25.05289

Setpoints 74&66 74&67 74&68 74&69 74&70 74&71 74&72 74&73

EUI 18.84156 19.12671 19.46808 19.89302 20.41507 21.06142 21.82168 22.61591

Setpoints 75&66 75&67 75&68 75&69 75&70 75&71 75&72 75&73 75&74
EUI 16.17537 16.45174 16.78144 17.18433 17.68205 18.28905 19.01197 19.85932 20.78974
Setpoints 76&66 76&67 76&68 76&69 76&70 76&71 76&72 76&73 76&74
EUI 13.47269 13.7406 14.06039 14.44831 14.92228 15.49893 16.18366 16.99402 17.92287
Setpoints 77&66 77&67 77&68 77&69 77&70 77&71 77&72 77&73 77&74
EUI 11.36324 11.62252 11.93348 12.30759 12.76552 13.31649 13.9662 14.73858 15.63866
Setpoints 78&66 78&67 78&68 78&69 78&70 78&71 78&72 78&73 78&74
EUI 9.464068 9.715562 10.01868 10.38111 10.82477 11.35517 11.98192 12.71359 13.57217

37
Table 12: Cooling and Heating EUI of the Second Floor

Figure 31: Cooling and Heating EUI of the Second Floor

Setpoints 70&66 70&67 70&68 70&69

Cooling Setpoint & Heating Setpoint
EUI 46.42825 47.02147 47.67669 48.38861

Setpoints 71&66 71&67 71&68 71&69 71&70

EUI 41.46739 42.02214 42.69629 43.4321 44.17936

Setpoints 72&66 72&67 72&68 72&69 72&70 72&71

EUI 36.22211 36.74645 37.37284 38.13227 38.94952 39.6691

Setpoints 73&66 73&67 73&68 73&69 73&70 73&71 73&72

EUI 32.05134 32.56145 33.15204 33.87095 34.702 35.60234 36.47633

Setpoints 74&66 74&67 74&68 74&69 74&70 74&71 74&72 74&73

EUI 28.00546 28.50704 29.08386 29.75712 30.55292 31.47034 32.45367 33.51541

Setpoints 75&66 75&67 75&68 75&69 75&70 75&71 75&72 75&73 75&74
EUI 24.63314 25.1251 25.69402 26.3404 27.08439 27.96925 28.98142 30.06554 31.31966
Setpoints 76&66 76&67 76&68 76&69 76&70 76&71 76&72 76&73 76&74
EUI 21.21289 21.69748 22.26069 22.89262 23.60621 24.43848 25.40731 26.52458 27.71921
Setpoints 77&66 77&67 77&68 77&69 77&70 77&71 77&72 77&73 77&74

38
EUI 18.47728 18.9548 19.50902 20.13181 20.82922 21.624 22.54903 23.61043 24.82979
Setpoints 78&66 78&67 78&68 78&69 78&70 78&71 78&72 78&73 78&74
EUI 16.00481 16.47678 17.02228 17.6373 18.32544 19.09906 19.97876 21.00799 22.16646
Table 13: Cooling and Heating EUI of the Third Floor

Figure 32: Cooling and Heating EUI of the Third Floor

The following points are listed here to explain the key findings of Table 10-13, Figure 29-32:
- For each floor, when the setpoints combination is 70&69, the cooling and heating EUI reaches
highest value. When cooling setpoint is 78 while heating is 66, the cooling and heating EUI have the lowest
value.
- -For each floor, the variation tendency of heating and cooling EUI in terms of changing setpoints
are similar. When cooling setpoint is constant while heating setpoint increases, the energy consumption will
increase. When heating setpoint remains unchanged and cooling setpoint increase, the energy consumption
will decrease.
- When the setpoint combination is constant, the third floor always has highest cooling and heating
EUI while the basement always has the lowest value. The higher the building level, the larger the cooling
and heating EUI.
- The heating and cooling EUI of the first floor are very close to values of the second floor.
- When cooling and heating set point temperature both increase, the cooling and heating EUI for each
floor moves closer and closers.

4.5 Zone-Level Energy Simulation Results

39
The zone-level energy simulation results are the most essential output for enabling a carbon-aware thermostat, because
the setpoint adjusting feedback is based on room level but room-level energy simulation is too complicate to generalize
to the entire building rooms or the campus building, on the other hand, the floor-level energy simulation is not
acceptable because it is too general for occupants to understand the carbon emission of the space they are currently in.
Thus, the energy consumption and carbon emission based on zone-level is the optimum simulation scale. Although
for generating carbon-aware thermostat, the simulation output for every thermal zone were required to export into
TrojanSense, this thesis only selected two thermal zones as examples to discuss and analyze. These two thermal zones
are zone 11 and zone 16, at the south of the second and third floor of Watt Hall, respectively. The table format is used
for changing data from simulation tool to user interface and the graphs indicate the differences in energy consumption
for different zones, which is the main reason to have zone-level rather than only floor-level energy simulation.
The following two figures (Figure 33 and 34) are total annual carbon emissions for zone 11 and zone 16. These results
were obtained through thermal zone splitting and energy simulation in EnergyPlus (via Honeybee plugin to
Grasshopper) and exported into excel, then converted energy consumption to carbon emissions. Similar to other graphs,
the results are based on different setpoint combinations to research how much energy consumption can result from a
determined setpoint combination.

Figure 33: Annual Carbon Emissions (pounds) for Zone 11 (2rd Floor South)

40

Figure 34: Annual Carbon Emissions (pounds) for Zone 16 (3
rd
Floor South)

The following points summarized key finding from Figure 33 and Figure 34:
- When the cooling set point temperature is 70℉ while heating set point temperature is 69℉, the total carbon emission
has the greatest value compared to other 65 setpoint combinations.
- When the cooling set point temperature is 78℉ while heating set point temperature is 66℉, the total carbon emission
has the lowest value compared to other 65 setpoint combinations.
- The variation trend in terms of setpoints changing are similar for two zones: When cooling set point temperature is
constant, and heating set point temperature gradually increases, the carbon emission increases. Conversely, when the
heating setpoint is constant while cooling setpoints increases gradually, the total carbon emission decrease.
- There is great difference of total carbon emission between two zones even though they are facing the same direction.
The multiple between zone 16 and zone 11 varies from 3.1 to 3.8.
The final output for raising occupants’ carbon awareness is telling them the percentage of annual carbon emissions
will be reduced or increased by adjusting current setpoints. As mentioned before, according to USC room temperature
policy, the current setpoints are 76℉for cooling and 68℉for heating. The carbon emission generated at this setpoint
combination is used as baseline value because it is the original carbon emission for this building without adjusting
setpoints. The following tables (Table 14-17) show the percentage of carbon saving or production after adjusting
setpoints for zone 11 and zone 16. The negative number means the carbon emission is reduced if change the setpoints
to the correspond value, comparing to the current carbon emission, and the positive number means the carbon emission
is increased.
Zone 11 - Assuming Current Cooling Setpoint is 76℉ and heating is 68℉
Setpoints 70&66 70&67 70&68 70&69
Cooling Setpoint & Heating Setpoint
Percentage 114.25 115.34 116.86 118.82

Setpoints 71&66 71&67 71&68 71&69 71&70

Percentage 91.17 92.16 93.50 95.35 97.72

Setpoints 72&66 72&67 72&68 72&69 72&70 72&71

41
Percentage 68.23 69.13 70.33 71.98 74.26 77.03

Setpoints 73&66 73&67 73&68 73&69 73&70 73&71 73&72

Percentage 48.75 49.57 50.66 52.15 54.18 56.87 60.02

Setpoints 74&66 74&67 74&68 74&69 74&70 74&71 74&72 74&73

Percentage 30.19 30.94 31.93 33.27 35.10 37.55 40.68 44.30

Setpoints 75&66 75&67 75&68 75&69 75&70 75&71 75&72 75&73 75&74
Percentage 14.22 14.90 15.81 17.03 18.69 20.92 23.78 27.31 31.49
Setpoints 76&66 76&67 76&68 76&69 76&70 76&71 76&72 76&73 76&74
Percentage -1.44 -0.85 0.00 1.11 2.62 4.68 7.31 10.58 14.50
Setpoints 77&66 77&67 77&68 77&69 77&70 77&71 77&72 77&73 77&74
Percentage -14.27 -13.76 -12.97 -11.94 -10.56 -8.67 -6.23 -3.17 0.47
Setpoints 78&66 78&67 78&68 78&69 78&70 78&71 78&72 78&73 78&74
Percentage -25.70 -25.24 -24.52 -23.56 -22.30 -20.57 -18.29 -15.43 -12.01
Table 14: Percentage of Carbon Increasing/Decreasing Compared to Current Carbon Emission of Zone11
(Assuming Current Cooling Setpoint is 76℉ and heating is 68℉)

Zone 16 - Assuming Current Cooling Setpoint is 76℉ and heating is 68℉
Setpoints 70&66 70&67 70&68 70&69
Cooling Setpoint & Heating Setpoint
Percentage 94.23 95.72 97.56 99.59

Setpoints 71&66 71&67 71&68 71&69 71&70

Percentage 75.05 76.39 78.15 80.26 82.56

Setpoints 72&66 72&67 72&68 72&69 72&70 72&71

Percentage 56.09 57.37 58.95 60.98 63.37 65.95

Setpoints 73&66 73&67 73&68 73&69 73&70 73&71 73&72

Percentage 39.74 40.96 42.47 44.31 46.63 49.28 52.14

Setpoints 74&66 74&67 74&68 74&69 74&70 74&71 74&72 74&73

Percentage 24.27 25.43 26.88 28.60 30.74 33.32 36.30 39.60

Setpoints 75&66 75&67 75&68 75&69 75&70 75&71 75&72 75&73 75&74
Percentage 10.75 11.87 13.25 14.93 16.92 19.32 22.16 25.52 29.25
Setpoints 76&66 76&67 76&68 76&69 76&70 76&71 76&72 76&73 76&74
Percentage -2.40 -1.33 0.00 1.62 3.53 5.78 8.43 11.64 15.32
Setpoints 77&66 77&67 77&68 77&69 77&70 77&71 77&72 77&73 77&74
Percentage -13.30 -12.27 -10.99 -9.43 -7.58 -5.40 -2.94 0.05 3.58
Setpoints 78&66 78&67 78&68 78&69 78&70 78&71 78&72 78&73 78&74
Percentage -22.92 -21.93 -20.70 -19.18 -17.39 -15.29 -12.91 -10.14 -6.82

42
Table 15: Percentage of Carbon Increasing/Decreasing Compared to Current Carbon Emission of Zone16
(Assuming Current Cooling Setpoint is 76℉ and heating is 68℉)
Although the room temperature policy for USC regulates that the cooling and heating setpoints are 76℉ and 68℉
respectively, according to the real-time temperature data for different rooms in Watt Hall (Figure 28), the average
temperature for zone 11 is 70℉ and 74℉ for zone 16, so the average temperature for zone11 and zone 16 of Watt Hall
is closer to 72℉ in cooling season. The following two tables (Table 16&17) show the percentage of carbon
increasing/decreasing to current carbon emission when the current cooling and heating setpoints are 72℉ and 68℉
respectively. These two tables are used as data transfer format from grasshopper to mobile app.
Zone 11 - Assuming Current Cooling Setpoint is 72℉ and heating is 68℉
Setpoints 70&66 70&67 70&68 70&69
Cooling Setpoint & Heating Setpoint
Percentage 25.78 26.42 27.32 28.47

Setpoints 71&66 71&67 71&68 71&69 71&70

Percentage 12.23 12.81 13.60 14.69 16.08

Setpoints 72&66 72&67 72&68 72&69 72&70 72&71

Percentage -1.23 -0.71 0.00 0.97 2.31 3.93

Setpoints 73&66 73&67 73&68 73&69 73&70 73&71 73&72

Percentage -12.67 -12.19 -11.55 -10.68 -9.48 -7.90 -6.06

Setpoints 74&66 74&67 74&68 74&69 74&70 74&71 74&72 74&73

Percentage -23.57 -23.13 -22.54 -21.76 -20.69 -19.25 -17.41 -15.28

Setpoints 75&66 75&67 75&68 75&69 75&70 75&71 75&72 75&73 75&74
Percentage -32.94 -32.55 -32.01 -31.29 -30.32 -29.01 -27.33 -25.26 -22.81
Setpoints 76&66 76&67 76&68 76&69 76&70 76&71 76&72 76&73 76&74
Percentage -42.14 -41.79 -41.29 -40.64 -39.75 -38.55 -37.00 -35.08 -32.78
Setpoints 77&66 77&67 77&68 77&69 77&70 77&71 77&72 77&73 77&74
Percentage -49.67 -49.37 -48.90 -48.30 -47.49 -46.38 -44.95 -43.15 -41.02
Setpoints 78&66 78&67 78&68 78&69 78&70 78&71 78&72 78&73 78&74
Percentage -56.38 -56.11 -55.68 -55.12 -54.38 -53.37 -52.03 -50.35 -48.34
Table 16: Percentage of Carbon Increasing/Decreasing Compared to Current Carbon Emission of Zone11
(Assuming Current Cooling Setpoint is 72℉ and heating is 68℉)

Zone 16 - Assuming Current Cooling Setpoint is 72℉ and heating is 68℉
Setpoints 70&66 70&67 70&68 70&69
Cooling Setpoint & Heating Setpoint
Percentage 21.71 22.69 23.90 25.23

Setpoints 71&66 71&67 71&68 71&69 71&70

Percentage 9.82 10.82 11.86 13.25 14.75

Setpoints 72&66 72&67 72&68 72&69 72&70 72&71

Percentage -1.90 -1.05 0.00 1.39 2.89 4.55

43
Setpoints 73&66 73&67 73&68 73&69 73&70 73&71 73&72

Percentage -12.03 -11.21 -10.22 -9.01 -7.50 -5.77 -3.92

Setpoints 74&66 74&67 74&68 74&69 74&70 74&71 74&72 74&73

Percentage -21.61 -20.84 -19.88 -18.74 -17.35 -15.67 -13.74 -11.62

Setpoints 75&66 75&67 75&68 75&69 75&70 75&71 75&72 75&73 75&74
Percentage -30.00 -29.26 -28.34 -27.23 -25.93 -24.37 -22.53 -20.37 -17.98
Setpoints 76&66 76&67 76&68 76&69 76&70 76&71 76&72 76&73 76&74
Percentage -38.15 -37.44 -36.55 -35.48 -34.23 -32.76 -31.04 -28.98 -26.62
Setpoints 77&66 77&67 77&68 77&69 77&70 77&71 77&72 77&73 77&74
Percentage -44.93 -44.24 -43.38 -42.35 -41.13 -39.72 -38.12 -36.19 -33.93
Setpoints 78&66 78&67 78&68 78&69 78&70 78&71 78&72 78&73 78&74
Percentage -50.92 -50.25 -49.43 -48.42 -47.24 -45.88 -44.33 -42.53 -40.51
Table 17: Percentage of Carbon Increasing/Decreasing Compared to Current Carbon Emission of Zone16
(Assuming Current Cooling Setpoint is 72℉ and heating is 68℉)
Table 14 to Table 17 shows the example of final table transferring format from simulation tools to the TrojanSense
project. By exporting these tables into mobile app or the TrojanSense web, users can understand how much percentage
of carbon emission will increase or decrease by the temperature adjustment.
4.6 Summary
This chapter illustrated the simulation results in terms of building level, floor level and zone level by graphs. The
results include total annually heating and cooling load (KBtu), total annually heating and cooling EUI (kBtu/sqft),
total annually carbon emission (pounds), total annually carbon emission intensity (pounds/sqft) and percentage of
carbon saving/causing compare to current carbon emissions (%) and all the results are based on 66 different setpoint
combinations. These figures and tables indicated that 1). Different floors or different zones will generate different
energy consumption. 2). Different cooling and heating setpoints combination will result in different energy
consumption. 3). Although the variation changes are similar for zones, floors and the building, the exact value for
heating and cooling load or EUI, or percentage of carbon saving/causing are significantly different from one thermal
zone to another, which is the main reason to have energy modeling in zone level. These findings will be discussed in
Chapter 5. Except indicating the zone differences, these results also demonstrated the possibility of generating data
transfer table from simulation software to mobile app via excel format, which bridge the gap between energy
simulation tools and the TrojanSense mobile app. By using these table, the carbon emission of each zones can be
presented to users directly, so users are able to understand the carbon impacts when they are adjusting setpoints on
TrojanSense and evaluate the importance of thermal comfort and energy efficiency, which might make contributions
to energy savings ultimately.

44
5. DISCUSSION
This chapter discussed the simulation results in chapter 4 by illustrating the tables into graphs or comparing different
results. The analysis follows the same structure, which is from entire building simulation results to the simulation
output of each zone. The purpose of investigating these results is demonstrate the significance of having zone level
energy simulation with different setpoint combinations while the results in chapter shows the feasibility of generating
data exchange format for carbon-aware thermostat.
5.1 Analysis of Building Level Energy Simulation Results
As table 9 and figure 28 show, obviously, different setpoints combinations can result in different heating and cooling
load. The highest annual cooling and heating load happens when setpoints are 70℉ and 69℉ respectively. When
cooling setpoint is 78℉and heating setpoint is 66℉, the cooling and heating load reaches the lowest value compared
to other 65 setpoints combinations and correspond heating and cooling load. The basic variation tendency can be
concluded from Table X: When cooling set point temperature is constant, and heating set point temperature gradually
increase, the total load increase. In the opposite, when the heating setpoint is constant while cooling setpoints increase
gradually, the total heating and cooling load decrease.
5.2 Analysis of Floor Level Energy Simulation Results
Figure 35 compared annual heating and cooling Energy Use Intensity (EUI) of each floor in Watt Hall. The EUI is the
total energy consumption of each floor divided by floor area, which is more appropriate for comparing energy
consumption than other parameters. Each point represents the heating and cooling EUI with a combination of setpoints.
The overall variation trend at different setpoint combinations are illustrated. Obviously, when setpoints combination
is in the same group, different floors have different heating and cooling energy use intensity. The third floor always
has the highest heating and cooling EUI while lowest heating and cooling EUI is always generated by basement. ℉The
higher the building level, the larger the cooling and heating EUI. The heating and cooling EUI of the first floor are
very close to the values of the second floor. When cooling or heating set point gets higher, the difference of total
heating and cooling EUI becomes smaller, which means the cooling and heating EUI for each floor moves closer and
closer. Although the values of heating and cooling EUI are different for different floors, the variation tendency are
similar for each floor.

45

Figure 35: Annual Heating and Cooling EUI (KBtu/sqft) Comparison Between Each Floor
The purpose of running energy simulation on floor level is to investigate if the same setpoints combination have the
same carbon impact for different floor. For instance, if the cooling setpoint is 78 and heating is 70, does this setpoints
combination always result in lower energy consumption than another setpoints (e.g. 77 for cooling and 76 for heating).
To solve this doubt, the setpoints combination are ranked in terms of value of heating and cooling EUI. Table 18 is
partially ranked setpoints combinations and corresponded heating and cooling EUI (from lowest value to highest
value). The table indicates that 1) same setpoints combination may be sorted into different level for different floors;
For example, as the table shows, the setpoints combination 77&66 results in the fourth lowest heating and cooling
EUI for basement level. However, for the other three ground floors, the same setpoint combination will lead to the
sixth, seventh and sixth lowest cooling and heating EUI respectively. 2) Two setpoints may have different carbon
impacts for different floor. For instance, the setpoints combination 78&70 will result in more heating and cooling EUI
than 77&66 for basement level. However, for the other three ground floor, the setpoints 78&77 can generate lower
heating and cooling EUI than setpoints 77&66.
Rank 1 2 3 4 5 6 7 8 9
Basement
78&66 78&67 78&68 77&66 78&69 77&67 77&68 78&70 76&66
6.136149 6.560845 7.106641 7.502325 7.778815 7.942156 8.502315 8.599986 9.028543
First Floor
78&66 78&67 78&68 78&69 78&70 77&66 78&71 77&67 77&68
8.823848 9.067968 9.367782 9.731988 10.17579 10.63491 10.7112 10.88722 11.19566
Second
Floor
78&66 78&67 78&68 78&69 78&70 78&71 77&66 77&67 77&68
9.464068 9.715562 10.01868 10.38111 10.82477 11.35517 11.36324 11.62252 11.93348
Third
Floor
78&66 78&67 78&68 78&69 78&70 77&66 77&67 78&71 77&68
16.00481 16.47678 17.02228 17.6373 18.32544 18.47728 18.9548 19.09906 19.50902

46
Table 18: Setpoints and EUI Sorted from Smallest to Largest Value
Figure 36 illustrates the sorted 66 setpoints combinations for each floor. The Y-coordinate is the ranked level from 1
to 66. 1 means the setpoints combination will lead to lowest heating and cooling EUI and 66 represents the highest
heating and cooling EUI. As the figure shows, the same setpoints combination may have different ranked level for
different floor. Although the first floor and second floor have the same floor area, the sorted results for these two
floors still have slightly differences.

Figure 36: Sorting Setpoints Based on How Much Cooling and Heating EUI Generating
5.3 Analysis of Zone Level Energy Simulation Results
Figure 37 is the dot plot to analyze and compare heating and cooling energy intensity between zone 11 and zone 16
with different 66 setpoint combinations. Since the floor area of these zones are different, it is necessary to use total
heating and cooling load divide floor area (pounds/sqft) to compare the energy consumption. The figure indicates that
the difference of heating and cooling energy intensity between two zones are fairly large. For the variation tendency,
it is much gentler for zone 11 than zone 16. Same with the building-level and floor-level simulation results, the largest
and lowest energy consumption result from two same setpoint combinations respectively. Figure 38 illustrate the
location of zone 11 and 16. Zone 11 is the south area of the second floor and zone 16 is the south area of the third
floor.

47

Figure 37: Annual Heating and Cooling Energy Intensity (pounds/sqft) from HVAC

Figure 38: Zone 11 (Left) and Zone 16 (Right)

The following figures show the percentage of carbon emission increasing/decreasing after adjusting setpoints for zone
11 and zone 16. The Y-coordinate includes all the possible setpoint combinations. One of these setpoint combination
is 76℉for cooling and 68℉for heating, which is the current setpoint and its percentage is 0. The X-coordinate shows
the percentage value. The positive number means the carbon emission will increase and negative number means carbon
emission will reduce. This figure indicates that most setpoint combinations will results in increasing carbon emission.
Moreover, same setpoint combination can have different effect on carbon impact for different thermal zones. For
instance, when cooling setpoint is 77℉ and heating set point is 73℉, for zone 11, the carbon emission is lower than
current carbon emission. However, for zone 16, there will be more carbon emission generated. This also demonstrate
the reason of generating zone-level energy simulation: because same setpoints combination may have different carbon
impacts for different zones.

48

Figure 39: Percentage of Carbon Saving/Causing Compared to Current Carbon Emission of Zone11
(Assuming Current Cooling Setpoint is 76℉ and heating is 68℉)

Figure 40: Percentage of Carbon Saving/Causing Compared to Current Carbon Emission of Zone16
(Assuming Current Cooling Setpoint is 76℉ and heating is 68℉)

49
The following two figures illustrate percentage of carbon emission increasing/decreasing with 66 setpoints
combinations of zone 11 and 16, which generated from table 16 and 17. Different with two figures above, the current
setpoints is assumed as 72℉ for cooling and 68℉ for heating because this is the actual average room temperature
monitored by sensors, which used as ultimate data exporting into TrojanSense app. Compared to the other two figures
above, there are much more setpoints combinations can lead to carbon saving if assuming current cooling setpoint is
72℉ instead of 76℉. These two figures also prove that same setpoints combination may result in different carbon
impacts (increase/decrease) for different thermal zones.

Figure 41: Percentage of Carbon Saving/Causing Compared to Current Carbon Emission of Zone11
(Assuming Current Cooling Setpoint is 72℉ and heating is 68℉)

50

Figure 42: Percentage of Carbon Saving/Causing Compared to Current Carbon Emission of Zone160
(Assuming Current Cooling Setpoint is 72℉ and heating is 68℉)
5.4 Summary

Figure 43: Same Setpoints can Result in Different Carbon Impacts for Different Zones

51
This chapter analyzed the results including building level, floor level and zone level simulation outputs. These results
demonstrate that 1) For entire building, each floor and zone, the variation tendency of simulation output are similar
with different setpoint combinations. When cooling setpoint is constant and heating setpoint gradually increase, the
total heating and cooling load will increase. However, when heating setpoint is constant and cooling setpoint gradually
increase, the total heating and cooling load will decrease.2) For different floors, the same setpoints may result in
different carbon impacts. For Watt Hall, the third floor always has the largest energy consumption while the basement
has the lowest energy consumption. The simulated output also indicated the first and second floor have the similar
energy consumption under different setpoint combinations. 4) It is necessary to have zone level building energy
modeling because same set point may have different impacts for different zones. Figure 43 illustrated this key finding.
For different two zones, for instance, zone 11 and zone 16, when current and user adjusted set point temperature are
same for different zones, the percentage of carbon saving/causing are be different as well, which is the benefit to have
the zone-level energy simulation. 5). The percentage of carbon saving/causing results, as the table format, will be
exported into TrojanSense for users to understand the carbon impacts resulted from different setpoints combination,
which bridge the gap between EnergyPlus and the TrojanSense mobile app.

52
6. CONCLUSION AND FUTURE WORK
6.1 Conclusions
The main contribution of this study is to express the limitation of current occupant-participatory temperature
adjustment approach in regard to lack of carbon awareness and carbon feedback, then to propose and develop
simulation-based methods to address it in a manner that could be scaled across a large set of campus buildings to
produce plasuable feedback which could be integrated into future upgrades to the current user interface of the
TrojanSense APP.
This thesis addressed a current limitation of current occupant-participatory temperature adjustment approach, where
users are unaware of the carbon impacts of both the current temperature setpoints (if they deviate from policy) as well
as the impacts of their desired adjustments (if any) to the current heating and cooling setpoint range.
To bridge the gap between occupants’ virtual temperature adjustment and energy efficiency, a new method called
carbon-aware thermostat was introduced. The carbon-aware thermostat can tell user how much carbon
increasing/decreasing will generate by their temperature adjustment, so users can evaluate current and desired
adjustments to temperature in context with carbon emissions. The carbon-aware thermostat can be achieved by
parametrically pre-running energy simulations in grasshopper and pre-calculation in excel based on all the possible
adjusted setpoints combinations.
This study simulated Watt Hall on the USC campus at three different levels of resolution for thermal zoning: building
level, floor level and zone level. The results indicate that although the annual energy consumption and carbon emission
have similar variation tendency, the different setpoints combination can lead to different energy consumption/carbon
emissions. For the zone-level building energy modeling, if the setpoint temperatures are same, different zone can have
different carbon impacts. Moreover, the carbon increase or decrease can be opposite with each other for different two
zones with same setpoints. These results demonstrated the significance of generating zone-level building energy
modeling rather than only floor level or building level for carbon-aware thermostat.
From an energy simulating point of view, this research tested the feasibility of zone-level energy modeling in
grasshopper associated with Honeybee and EnergyPlus and introduced the simulation parameters set-up process.
The research also investigated the approach to transfer simulation output from grasshopper to the mobile app and
introduced the table format. This contribution allows “real-time” simulation feedback from the perspective of users.
6.2 Future Work
Since this study is the initial step of energy modeling for TrojanSense project, the scope of this research is not enough
for the entire project. Many different adaptations and simulations have been left for the future due to lack of time. For
this study, future work concerns more energy simulations, more target buildings for analysis, results validation, new
proposal for data transfer and some improvements for current study.
6.2.1 Campus Building Energy Modeling
As mentioned before, the energy modeling for this project is aims to generating campus building energy model
ultimately. Currently, this study only tested the feasibility of creating real-time carbon aware model for one building.
In order to have larger scope of the project, the energy modeling for other buildings in USC campus are necessary for
this project. In this thesis, one possible method for campus building energy modeling was proposed and discussed,
which requires more building information and divide buildings into different groups in terms of their ArchTypes, then
run energy simulation for each group (see details in section 2.3 and 3.5), and it can be evaluated in the future to test
its feasibility. There can be some other solutions for large district energy modeling, such as using city modeling
software.
6.2.2 Building Retrofit Strategies

53
As a carbon-aware education tool, it is not enough to only provide building energy simulation in current situation.
Some possible retrofit solutions and their carbon impacts can be introduced in the future, so occupants are able to
consider how to make the building more sustainable and environmental. To apply this idea, the energy simulation
process should not only focus on current building information, other possible retrofit strategies can be applied to
replace the existing building parameters. For instance, the current building envelop for Watt Hall is concrete mass
wall without insulation, so one of the retrofit strategies can be adding insulation for the envelope. Some other strategies,
such as changing current double-glazing glass to triple glazing, or changing current VAV HVAC system. Every retrofit
strategy and its impacts are pre-simulated in tools and export to excel as table format. Concerning the user’s mobile
app, there will be another interface for the building retrofitting. In this interface, various retrofit strategies are listed
and described, occupants can select any one of them to see the energy consumption and carbon impacts if applying
this strategy.
6.2.3 Validation
Results validation is an essential part for energy simulation. The common method for simulation results validation is
comparing the annual simulation results to the utility bills to evaluate the difference. However, it is not suitable for
this study due to several reasons: 1. Most simulation output is based on zone level. It is too complicate to conduct
energy consumption of a zone from the building utility bills. 2. The results simulated the energy consumption with
different heating and cooling setpoints while the actual energy consumption is only determined by the specific
setpoints. 3. The weather data used for energy simulation is based on historical record instead of real-time data, which
results in difference between simulation output and actual energy consumption. Finding the appropriate approach for
results validation of this research is left for future work.
In addition to validate the accuracy of simulation results, the effect of carbon-aware thermostat can be evaluated for
future work. This study only discussed the method of generating real-time simulation feedback for user-engaged
temperature management, however, whether this carbon-aware thermostat can make contribution for energy/carbon
saving or not, whether users do consider the energy efficiency when being told the percentage of carbon
increasing/decreasing, still need more time and work to collect users’ feedback to validate it. This validation can be
achieved by comparing average virtual setpoints before and after having carbon-aware thermostat for TrojanSense.
6.2.4 Data Transfer
As mentioned before, the simulation is completed in grasshopper while the data format, which used for exporting into
mobile app, should be exported and processed in excel (csv file) and then be imported in TrojanSense. Although the
simulation results can be exported into csv file directly, however, it requires manually data processing work to pick
useful results and associate it with timeline. Currently this study only includes monthly and annual simulation output,
which took little time for data cleaning. However, in the future work. The study scope will extent to entire campus
including several buildings with lots of thermal zones, and the simulation results might be detailed to every day or
even every hour, so it will be helpful to generate a program or find other solutions for data transfer and cleaning in
grasshopper.
6.2.4 Other Improvements
In addition to adding campus building modeling, building retrofit strategies and results validation, there are some other
improvements left for future study, including refine current setpoints, add daily and hourly energy simulation and
enable more occupants understand the output.
The current setpoints for most USC buildings is 76℉for cooling and 68℉for heating according to USC facility
management service (FMS). However, this setpoint combination is unable to reflect the real-time room temperature,
which may confuse occupants to understand the actual energy saving/causing. For instance, in the cooling season, a
room with the constant set point (76℉for cooling), the actual room temperature will fluctuate a few degrees.

54
So far it is possible to complete daily and hourly energy simulation in simulation tools, however, since the data for
daily and hourly energy simulation is much more compared to monthly data, therefore, it has difficulty to transfer
these data from excel format to mobile app. If this technology can be achieved, uses are able to select the energy
consumption at a specific day or time to see how much energy consumption will generate in the specific zone.
Another improvement is avoiding academic terms and using more understandable message to explain the energy
consumption and carbon emission for occupants. For instance, BTU as the metric for energy consumption, is not
understandable for most users who without building science background. Thus, it is necessary to explain the terms in
an accessible way.

55
APPENDIX
Zone 11 – 2
nd
Floor South -Watt Hall Heating and Cooling Load (KBtu)

Cooling Heating Cooling Heating Cooling Heating Cooling Heating Cooling Heating

70 66 70 67 70 68 70 69 71 66
Jan 7038.113 364.8465 7061.608 520.0504 7088.823 720.0363 7118.026 938.3944 6405.817 320.902

7402.959344 7581.658798 7808.859367 8056.419979 6726.719041
Feb 5520.111 209.4843 5541.961 347.57 5567.481 516.3516 5595.956 700.5912 5009.248 174.4731

5729.595528 5889.531447 6083.832885 6296.547119 5183.720825
Mar 5939.858 118.7724 5959.646 206.9842 5985.554 340.2017 6015.377 498.1372 5246.864 95.06552

6058.630621 6166.630012 6325.7556 6513.514528 5341.929383
Apr 5606.445 84.56998 5612.608 159.5425 5622.803 269.3809 5637.695 436.0889 4817.007 69.18896

5691.01477 5772.150864 5892.183763 6073.784098 4886.196237
May 8356.453 4.575608 8362.872 19.48623 8376.208 55.27966 8400.769 141.6783 6877.812 3.502797

8361.028353 8382.358103 8431.487975 8542.447082 6881.314322
Jun 8267.952 0 8268.246 0 8269.944 2.289135 8275.968 14.84426 6664.125 0

8267.951736 8268.245827 8272.233002 8290.812324 6664.124595
Jul 15504.53 0 15504.53 0 15504.67 0 15505.54 0 13678.88 0

15504.53459 15504.53093 15504.66639 15505.53721 13678.88261
Aug 22323.71 0 22323.7 0 22323.93 0 22324.82 0.066282 20505.29 0

56

22323.71035 22323.69853 22323.93248 22324.88751 20505.28972
Sep 23723.35 0 23723.41 0 23723.98 0 23726.45 0 21516.4 0

23723.34643 23723.40867 23723.97897 23726.44658 21516.40296
Oct 21922.84 0 21926.77 0 21933.93 3.111242 21948.9 16.13737 19673.22 0

21922.83684 21926.76868 21937.03735 21965.03839 19673.22131
Nov 14485.52 21.29235 14501.04 45.20389 14522.69 93.11547 14555.14 185.5408 13139.08 14.89603

14506.81719 14546.24885 14615.80995 14740.68061 13153.97786
Dec 11131.24 291.8907 11159.64 438.285 11196.73 636.5172 11238.75 857.3642 10185.55 258.6043

11423.12977 11597.92781 11833.24693 12096.11727 10444.15383
Annual 149820.1 1095.432 149946 1737.122 150116.7 2636.283 150343.4 3788.843 133719.3 936.6327
TOTAL 150915.5555 151683.1585 152753.0247 154132.2327 134655.9327
Table 19: Monthly and Annually Cooling and Heating Load of Zone 11 (2rd Floor South) at different setpoints combinations

57
Zone 16 - 3rd Floor South - Watt Hall Heating and Cooling Load (KBtu)

Cooling Heating Cooling Heating Cooling Heating Cooling Heating Cooling Heating

70 66 70 67 70 68 70 69 71 66
Jan 12350.12 1758.726 12379.17 2104.343 12410.88 2502.599 12445.81 2884.461 11327.49 1678.081

14108.84831 14483.51156 14913.48382 15330.27515 13005.56816
Feb 10202.75 1241.527 10227.23 1534.313 10255.28 1878.379 10284.83 2207.748 9378.825 1181.641

11444.27479 11761.54024 12133.65683 12492.57653 10560.46682
Mar 10876.59 879.3988 10901.96 1148.463 10930.52 1455.821 10966.52 1793.209 9810.091 822.6186

11755.99169 12050.42656 12386.33979 12759.72714 10632.70976
Apr 9383.846 478.5695 9397.589 728.2197 9415.473 1039.066 9435.786 1391.732 8148.619 440.0409

9862.415957 10125.80866 10454.53883 10827.51849 8588.659701
May 13382.58 106.4408 13394.19 220.318 13413.61 375.5022 13443.24 578.2829 10986.64 86.90357

13489.01997 13614.51109 13789.11464 14021.51801 11073.54781
Jun 13268.1 6.297675 13269.13 24.49832 13272.51 73.57713 13283.35 149.9129 10703.62 4.675184

13274.39372 13293.6327 13346.08511 13433.26635 10708.29234
Jul 24699.86 0 24699.85 0 24700.17 3.602519 24702.33 13.45876 22035.95 0

24699.85729 24699.84566 24703.77235 24715.793 22035.94615
Aug 35290.98 0 35290.96 0 35291.31 2.669259 35292.1 6.154054 32645.96 0

35290.97566 35290.95551 35293.97545 35298.2515 32645.96147

58
Sep 36225.54 0 36225.73 0.895162 36227.78 12.40899 36231.06 29.59404 33123.27 0

36225.53732 36226.62724 36240.19209 36260.65208 33123.26769
Oct 34687.73 51.12213 34696.81 89.0842 34710.6 156.0013 34733.64 254.1545 31645.31 44.10231

34738.85454 34785.8932 34866.60318 34987.7942 31689.41603
Nov 24099.77 354.5336 24127.15 531.964 24159.18 752.584 24198.43 1000.274 22023.04 319.8716

24454.30371 24659.11832 24911.76162 25198.70773 22342.9156
Dec 19429.89 1490.495 19463.23 1833.006 19500.91 2235.097 19543.64 2640.616 17996.63 1416.078

20920.38612 21296.23176 21736.00262 22184.25858 19412.70368
Annual 473242.2 6367.111 475064.9 8215.104 477327.7 10487.31 479886.8 12949.6 426232.2 5994.012
TOTAL 479609.332 483279.9732 487815.05 492836.4189 432226.2067
Table 20: Monthly and Annually Cooling and Heating Load of Zone 16 (3rd Floor South) at different setpoints combination.

59
REFERENCE
AASHE, the Association for the Advancement of Sustainability in Higher Education. (n.d.). Retrieved September 20,
2018, from https://www.aashe.org/

Barr, Michael. Pulse Width Modulation, Embedded Systems Programming, September 2001, p. 103
University of Southern California. (2017, April 27). Retrieved September 24, 2018, from
https://academicsenate.usc.edu/usc-sustainability-2030-strategy-april-27-2017/

University of Southern California. (n.d.) Retrieved September 10, 2018, from
https://facilities.usc.edu/leftsidebar.asp?ItemID=230
Zhang, L., A. H. Lam, and D. Wang. (2014). Strategy-proof Thermal Comfort Voting in
Buildings. BuildSys 2014 (pp. 160–163). New York, NY, USA: ACM.
http://doi.org/10.1145/2676061.2674074
Erickson, V. and A. Cerpa. (2012). Thermovote: Participatory Sensing for Efficient Building
HVAC Conditioning, In Proceedings of the Fourth ACM Workshop on Embedded Sensing
Systems for Energy-Efficiency in Buildings (BuildSys '12). ACM, New York, NY, USA, 9-
16. DOI=http://dx.doi.org/10.1145/2422531.2422534
Trojan Sense (2018). Retrieved Novemver 20, 2018, from http://trojansense.usc.edu/about/
Suesser, T. and Dogan, T. (2017). Campus Energy Model: Using a Semi-Automated Workflow to Build Spatially
Resolved Campus Building Energy Models for Climate Change and Net-Zero Scenario Evaluation. The 15
th
IBPSA
Conference. San Francisco, USA.
Sanguinetti, A., Pritoni, M., Salmon, K., & Morejohn, J. (2016). TherMOOstat: Occupant Feedback to Improve
Comfort and Efficiency on a University Campus. ACEEE Summer Study on Energy Efficiency in Building
Conference. Asilomar, USA.
Understanding Model Predictive Control. (n.d.). Retrieved October 30, 2018, from
https://www.mathworks.com/videos/series/understanding-model-predictive-control.html
Walter, T. G., Kwok, A.G., Stein, B., Reynolds, J.S. (2014) Mechanical and Electrical Equioment for Buildings 12
th

Edition. New York, NY: J. Wiley and Sons.
Pritoni, M., Salmon, K., Sanguinetti, A., Morejohn, J., & Modera, M. (2017). Occupant thermal feedback for improved
efficiency in university buildings. Energy and Buildings, 144, 241-250.
Balaji, B., H. Teraoka, R. Gupta, and Y. Agarwal. 2013. Zonepac: Zonal power estimation and control via hvac
metering and occupant feedback. 2013 Proceedings of the 5
th
ACM Workshop on Embedded Systems for Energy-
Efficient Buildings. ACM, 2013.
Dong, Z. Exploring Participatory Sensing and the Internet of Things to Develop Data-Driven Thermal Comfort
Models on the USC Campus. 2018.
Reinhart, C.R., & Dacila, C. C. (2016). Urban building energy modeling – A review of a nascent field. Building and
Environment, 97, 196-202.
Chen, Y., Hong, T., & Piette, M. A. (2017). Automatic generation and simulation of urban building energy models
based on city datasets for city-scale building retrofit analysis. Applied Energy, 206, 323-335.

60
Davila, C. C. (2017). Building archetype calibration for effective urban building energy modeling.
Reinhart, C. R., Dogan, T., Jakubiec, J. A., Rakha, T., & Sang, A. (2013) UMI – An urban simulation environment
for building energy use, daylighting and walkability. Proceedings of Building Simulation. Chambery, France.
Harvard University Temperature Policy. (2015, December 08). Retrieved September 12, 2018, from
https://green.harvard.edu/temperature-policy
Rice University Building Temperature Policy. (n.d.). Retrieved September 12, 2018, from
https://sustainability.rice.edu/building-temperature-policy
Indoor Space Temperature Guidelines. (n.d.). Retrieved September 12, 2018, from
https://policylibrary.columbia.edu/indoor-space-temperature-guidelines
Bentley University Temperature Policy. (2016). Retrieved September 12, 2018, from
https://www.bentley.edu/files/2016/08/11/2016%20Bentley%20University%20Temperature%20Policy.pdf
Tufts University Temperature Policy. (n.d.). Retrieved September 12, 2018, from
https://sustainability.tufts.edu/resources/emissions/tufts-university-temperature-policy/

Abstract (if available)

Abstract The current control methods of HVAC systems depend on static temperature set points based on standards such as ASHARE Standard 55, rather than the comfort requriements of the actual building occupants in a space, often leading to unnecessary over-heating or over-cooling. These control methods in turn result in unnecessary energy consumption and contribute to higher energy costs and carbon emissions. Although some researchers demonstrated that occupant- participatory temperature adjustment approach, i.e., incorporating users’ thermal comfort feedback via mobile app or web, can reduce energy consumption, those studies do not place comfort feedback and associated thermal changes in context with changes in carbon emissions. ❧ The aim of this thesis is to develop a simulation-based approach to generate the necessary building performance data to support the development of a carbon-aware virtual thermostat for mobile devices to enable building occupants to better-understand the impact of temperature setpoint changes on building energy use and associated carbon emissions. TrojanSense is a project encouraging students, faculty and staff to make changes to the indoor set point temperature in University of Southern California (USC) buildings. Watt Hall, as the main building housing the architecture school at USC, was chosen as the sample building. The building model was generated in Rhino and Grasshopper, then setting up all the parameters and running energy simulations in Honeybee and EnergyPlus. The simulation output is the data exchange format with all possible set point temperature combinations. By exporting this pre-calculated simulation results into TrojanSense mobile app or web, the occupant can look up how much energy consumption and carbon emission could be saved or wasted in real time when virtually adjusting the setpoint, which raises occupants’ carbon awareness and encourage them to pay more attention to sustainability.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Occupant-aware energy management: energy saving and comfort outcomes achievable through application of cooling setpoint adjustments

PDF

Exploring participatory sensing and the Internet of things to evaluate temperature setpoint policy and potential of overheating/overcooling of spaces on the USC campus

PDF

Developing environmental controls using a data-driven approach for enhancing environmental comfort and energy performance

PDF

Enhancing thermal comfort: data-driven approach to control air temperature based on facial skin temperature

PDF

Multi-occupancy environmental control for smart connected communities

PDF

Energy use intensity estimation method based on building façade features by using regression models

PDF

Post occupancy energy evaluation of Ronald Tutor Hall using eQUEST; computer based simulation of existing building and comparison of data

PDF

Environmentally responsive buildings: multi-objective optimization workflow for daylight and thermal quality

PDF

Dynamic shading and glazing technologies: improve energy, visual, and thermal performance

PDF

Natural ventilation in tall buildings: development of design guidelines based on climate and building height

PDF

The intelligent control strategy of kinetic façades for daylight and energy performance: evaluating the daylight effect of adaptive systems based on parametric workflow

PDF

A simplified building energy simulation tool: material and environmental properties effects on HVAC performance

PDF

Night flushing and thermal mass: maximizing natural ventilation for energy conservation through architectural features

PDF

Adaptive façade controls: A methodology based on occupant visual comfort preferences and cluster analysis

PDF

Developing a data-driven model of overall thermal sensation based on the use of human physiological information in a built environment

PDF

Impact of occupants in building performance: extracting information from building data

PDF

Enhancing thermal comfort: air temperature control based on human facial skin temperature

PDF

A BIM-based visualization tool for facilities management: fault detection through integrating real-time sensor data into BIM

PDF

Energy performance of different building forms: HEED simulations of equivalent massing models in diverse building surface aspect ratios and locations in the US

PDF

Nursing comfort study: are hospitals performing for nurses? IEQ analysis of working conditions inside hospitals

Asset Metadata

Creator Zhang, Danyang (author)

Core Title Real-time simulation-based feedback on carbon impacts for user-engaged temperature management

School School of Architecture

Degree Master of Building Science

Degree Program Building Science

Publication Date 07/09/2019

Defense Date 05/06/2019

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag building energy modeling,carbon emission,OAI-PMH Harvest,thermal comfort

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Konis, Kyle (committee chair), Noble, Douglas (committee member), Schiler, Marc (committee member)

Creator Email ddyang27@gmail.com,zhan057@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-181870

Unique identifier UC11662504

Identifier etd-ZhangDanya-7536.pdf (filename),usctheses-c89-181870 (legacy record id)

Legacy Identifier etd-ZhangDanya-7536.pdf

Dmrecord 181870

Document Type Thesis

Format application/pdf (imt)

Rights Zhang, Danyang

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA