Sustainable Human-Building Ecosystems: Yimin Zhu, Khee Poh Lam, Yong Tao

Sustainable
Human–Building
Ecosystems
Edited by Yimin Zhu, Ph.D.
Khee Poh Lam, Ph.D.
Yong Tao, Ph.D.
SUSTAINABLE
HUMAN–BUILDING
ECOSYSTEMS
SELECTED PAPERS FROM THE FIRST INTERNATIONAL SYMPOSIUM
ON SUSTAINABLE HUMAN–BUILDING ECOSYSTEMS
October 5-6, 2015

Pittsburgh, Pennsylvania
SPONSORED BY
Research Coordination Network on Sustainable Human–Building

Ecosystems of the University of North Texas
The Architectural Engineering Institute

of the American Society of Civil Engineers
EDITED BY
Yimin Zhu, Ph.D.
Khee Poh Lam, Ph.D.
Yong Tao, Ph.D.
Published by the American Society of Civil Engineers

Published by American Society of Civil Engineers
1801 Alexander Bell Drive
Reston, Virginia, 20191-4382
www.asce.org/publications | ascelibrary.org
Any statements expressed in these materials are those of the individual authors and do not necessarily
represent the views of ASCE, which takes no responsibility for any statement made herein. No reference
made in this publication to any specific method, product, process, or service constitutes or implies an
endorsement, recommendation, or warranty thereof by ASCE. The materials are for general information
only and do not represent a standard of ASCE, nor are they intended as a reference in purchase
specifications, contracts, regulations, statutes, or any other legal document. ASCE makes no
representation or warranty of any kind, whether express or implied, concerning the accuracy,
completeness, suitability, or utility of any information, apparatus, product, or process discussed in this
publication, and assumes no liability therefor. The information contained in these materials should not be
used without first securing competent advice with respect to its suitability for any general or specific
application. Anyone utilizing such information assumes all liability arising from such use, including but
not limited to infringement of any patent or patents.
ASCE and American Society of Civil Engineers—Registered in U.S. Patent and Trademark Office.
Photocopies and permissions. Permission to photocopy or reproduce material from ASCE publications
can be requested by sending an e-mail to permissions@asce.org or by locating a title in ASCE's Civil
Engineering Database (http://cedb.asce.org) or ASCE Library (http://ascelibrary.org) and using the
“Permissions” link.
Errata: Errata, if any, can be found at http://dx.doi.org/10.1061/9780784479681
Copyright © 2015 by the American Society of Civil Engineers.

All Rights Reserved.
ISBN 978-0-7844-7968-1 (PDF)
Manufactured in the United States of America.
Sustainable Human–Building Ecosystems iii
Preface
Welcome to the First International Symposium on Sustainable Human–Building

Ecosystems!
For decades, a significant amount of research effort has been directed towards
developing engineering, economic, and policy approaches to better design and
maintain energy efficient buildings. Until recently, many have realized that occupant
behavior is another missing piece of the equation. As the research community is
quickly growing, we need a venue where researchers can share cutting edge findings
in the integration of human behavioral, social and economic sciences with building
design, engineering and metrology for better understanding building energy
performance, environmental impacts and occupant comfort.
We received over 60 abstracts and 25 high quality papers have been accepted and
included in the proceedings. The abstracts and final papers were peer-reviewed. The
proceedings cover a wide range of topics, generally classified into three subject areas,
namely, occupant behavior modeling and analysis, thermal comfort prediction and
analysis, and innovative design, planning and policies for building energy efficiency.
The proceedings also include a keynote presentation paper on human ecology and
building science.
On behalf of the organizing committee, we would like to thank the School of

Architecture at Carnegie Mellon University for hosting the symposium, our sponsors
Autodesk Inc. and Cristal Global Engineering for their generous financial support,
and reviewers for their great contribution to the review process. We also want to
recognize the Predictive Modeling Network for Sustainable Human-Building
Ecosystems (SHBE), a Research Coordination Network (RCN) for Science,
Engineering and Education for Sustainability (SEES) funded by US National Science
Foundation (NSF) for establishing the foundation for this symposium, and
contributions from the participants of Annex 66: Definition and Simulation of
Occupant Behavior in Buildings under the auspices of the International Energy
Agency’s (IEA) Energy in Buildings and Communities (EBC) Programme.
© ASCE
Sustainable Human–Building Ecosystems iv
Contents
Featured Paper
Human Ecology and Building Science: A Necessary Synthesis .............................. 1

Ardeshir Mahdavi
Occupant Behavior Modeling and Analysis
Occupant Behaviors and Energy Use: Creating High-Performance People for

High-Performance Buildings ................................................................................... 18
Julia K. Day
Participatory Energy Management in Building Networks ................................... 27

Mina Rahimian, Daniel Cardoso-Llach, and Lisa Domenica Iulo
One Size Does Not Fit All: Eco-Feedback Programs Require

Tailored Feedback .................................................................................................... 36
Ardalan Khosrowpour and John E. Taylor
Development of Non-Intrusive Occupant Load Monitoring (NIOLM) in

Commercial Buildings: Assessing Occupants’ Energy-Use Behavior at
Entry and Departure Events.................................................................................... 44
Hamed Nabizadeh Rafsanjani, Changbum R. Ahn, and Mahmoud Alahmad
Default Conditions: A Reason for Design to Integrate Human Factors .............. 54

Arsalan Heydarian, Joao P. Carneiro, Evangelos Pantazis, David Gerber, and
Burcin Becerik-Gerber
Assessing Energy Strategies in Active Buildings Considering

Human Behaviour ..................................................................................................... 63
Ayesha Kashif, Stephane Ploix, and Julie Dugdale
Estimating Occupancy in an Office Setting............................................................ 72

Manar Amayri, Stephane Ploix, and Sanghamitra Bandyopadhyay
Event-Based Parallel Simulation with a Sensing System for Occupant

Distribution Estimation in the Whole Building Scale ........................................... 81
Zhenning Lang and Qing-Shan Jia
© ASCE
Sustainable Human–Building Ecosystems v
Effects of Variant Occupancy Transitions on the Energy Implications of

Setpoint/Setback Control Policies ........................................................................... 90
Zheng Yang, Ali Ghahramani, and Burcin Becerik
Thermal Comfort Prediction and Analysis
A Study of Time-Dependent Variations in Personal Thermal Comfort via a

Dynamic Bayesian Network ..................................................................................... 99
Ali Ghahramani, Chao Tang, Zheng Yang, and Burcin Becerik-Gerber
Occupant Individual Thermal Comfort Data Analysis in an Office .................. 108

Jie Zhao, Khee Poh Lam, Vivian Loftness, and B. Erik Ydstie
Facial Skin Temperature as a Proactive Variable in a Building Thermal

Comfort Control System ........................................................................................ 117
Bo Yi and Joon-Ho Choi
Advancing Occupant-Centered Performance Simulation Metrics Linking

Commercial Environmental Quality to Health, Behavior, and Productivity ... 126
M. Azarbayjani, D. Brentrup, and R. Cox
Direct Measurement of Occupants’ Skin Temperature and Human

Thermal Comfort Sensation for Building Comfort Control ............................... 141
Pooya Sharifani, Suraj Talele, Junghyun Mun, and Yong Tao
Innovative Planning, Design, and Policies for Building Energy Efficiency
Incorporation of Future Building Operating Conditions into the Modeling

of Building–Microclimate Interaction: A Feasibility Approach ........................ 150
Kelly Kalvelage, Ulrike Passe, Caroline Krejci, and Michael C. Dorneich
Measuring the Effectiveness of an Immersive Virtual Environment for the

Modeling and Prediction of Occupant Behavior ................................................. 159
Sanaz Saeidi, Tracy Rizzuto, Yimin Zhu, and Robert Kooima
Integrated Project Delivery and Total Building Automation for the Nearly
Net-Zero-Energy Q1 ThyssenKrupp Headquarters............................................ 168
Thomas Spiegelhalter
Green Building Design as If People Mattered ...................................................... 176

Maryam H. Kashani, Lyn Bartram, and Robert Woodbury
Integration of QFD and Utility Theory to Improve End-User Satisfaction

in the Design of High-Performance Buildings...................................................... 185
Ehsan Mostavi, Somayeh Asadi, Ebrahim Karan, and Djamel Boussaa
© ASCE
Sustainable Human–Building Ecosystems vi
The Power of Data Visualization: A Prototype Energy Performance Map

for a University Campus ........................................................................................ 194
Khaled A. Tarabieh, Islam O. Elnabarawy, Islam A. Mashaly, and
Yussra M. Rashed
Using Relationship Mapping to Understand Sustainable Housing

Stakeholders’ Actions ............................................................................................. 204
S. Zedan and W. Miller
The Weatherization Assistance Program: Social Policy or Energy Policy

and Why It Matters ................................................................................................ 214
J. N. Terman
Towards Multi-Objective Optimization for Sustainable Buildings with

Both Quantifiable and Non-Quantifiable Design Objectives.............................. 223
W. Yan, M. Rahmani Asl, Z. Su, and J. Altabtabai
Inequality as a Barrier to Green Building Policy Adoptions in Cities............... 231

Aaron Deslatte, Kathryn Wassel, and Richard C. Feiock
© ASCE
Sustainable Human–Building Ecosystems 1
Human Ecology and Building Science: A Necessary Synthesis
Ardeshir Mahdavi
Department of Building Physics and Building Ecology,

TU Wien, Karlsplatz 13, 1040 Vienna, Austria. E-mail: bpi@tuwien.ac.at
INTRODUCTION
Buildings represent a major class of interventions by people in the

environment, resulting in resource depletion, environmental emissions, and waste
generation. To date, building-related interventions, together with those involving
industry and transportation, have comprised the planet’s capacity to support the
sustenance of future generations. This circumstance has raised the awareness
regarding the importance of sustainability in the building sector. However, as in
many other environmentally relevant domains, the theory (how the built environment
should be) and practice (how the built environment actually is) have not converged.
If we are to take the idea of sustainable buildings seriously, we need to approach it
critically in the context of the complex and consequential relationships involving
people, buildings, and environment. Why do we erect buildings? How can we
measure their effectiveness in meeting people's requirements? How can we assess
their ecological implications?
To address these questions this paper offers a broad and critical framework.
Toward this end, section 2 of the paper entails a brief introduction to "Human
Ecology" as a fruitful conceptual framework for the discussion of interrelationships
between people and their surrounding built environment. Section 3 is dedicated to a
number of essential background or boundary conditions necessary for a meaningful
discussion of sustainability in the building sector. These include, amongst other
topics, population growth and "life style". Section 4 is concerned with the indoor
environmental (especially thermal) requirements of building users. Section 5
addresses people's passive and active influences on buildings' indoor climate and
environmental performance. Section 6 summarizes the paper's conclusions.
HUMAN ECOLOGY
As a discipline, ecology deals with the relationships between organisms and
their surrounding world. Accordingly, human ecology may be simply defined as the
ecology of the Homo sapiens. There are multiple traditions in human ecology. For
the purpose of the present discussion, we consider the "Vienna School of Human
Ecology" (Knötig ,1992a, 1992b; Mahdavi, 1996a) and focus on a couple of its
essential concepts. Construction and operation of buildings and related artifacts may
be viewed as an integral part of the totality of largely regulatory operations initiated
by human beings as they interact with their surrounding world. Human ecology
offers a useful way of thinking about these interactions via a number of high-level
yet versatile concepts. Thereby, a central pair of concepts involves:
i. the human beings’ ecological potency;
ii. the surrounding world’s ecological valency (Knötig, 1992a; Mahdavi,
1996b).
© ASCE
Ecological potency refers to the human repertoire of means to deal (cope,

interact) with the surrounding world. Ecological valency denotes the totality of that
surrounding world’s characteristics (resources, possibilities, opportunities,
challenges, risks, hazards, etc.) as it relates to, confronts, or accommodates people's
ecological potency. Coined initially by Uexküll (1920), the concept of ecological
valency is akin to the Gibson's concept of affordance (Gibson, 1977, 1979).
Given this conceptual framework, the main consideration in human ecology
pertaints to the complex and dynamic relationships between the ecological potency
of human beings and the ecological valency of their surrounding world. We can thus
broadly characterize the entire building construction and operation endeavor in
human ecological terms: Buildings are mainly constructed and maintained with the
(implicit or explicit) intention to favorably influence the relationship between
people's ecological potency and the ecological valency of their surrounding world.
Such an intention expresses itself, for example, in the "shelter function" of the
vernacular architecture (Mahdavi, 1996c, 1989). In contemporary building delivery
processes, this intention is often expressed explicitly and formally, for example when
desirable indoor environmental conditions are specifically defined and are expected
to be maintained in the course of building operation. Provision of desirable
conditions for the building users, or in other words, maintaining a high degree of
"habitability", may be thus seen as the central utility of buildings. The challenge is to
realize habitability with a minimum on resource depletion and environmental impact.
Human ecology's concepts are also relevant to the evaluation of the
habitability of the built environment. Specifically, a second pair of concepts should
be mentioned, which concerns distinct aspects of the relationships between people
and their surroundings. Thereby a high-level distinction is made between the
material-energetic and information-related aspects of these relationships (Knötig,
1992a; Mahdavi, 1996a, 1992). These two aspects can be assigned to every entity,
state, and process. The material-energetic aspect refers to the assumption that
nothing exists unless some amount of matter or energy is involved. The information-
related aspect refers to the assumption that matter and energy have a certain
distribution in space and time, which can be represented in terms of a structure. An
information content can be correlated with this structure.
To measure the habitability of the built environment we cannot disregard
people's subjective experiences and opinions. Subjective evaluation processes of the
built environment arguably involve both material-energetic and the information-
related aspects of the relationships between inhabitants and the built environment. A
common approach to "operationalize" such evaluation processes in planning and
operating involves the use of "psycho-physical" scales. The idea is that exposure to
various levels of physical (material-energetic) stimuli translates – in a more or less
predictable way – into corresponding subjective experiences. For example, exposure
to increasing levels of sound intensity is said to result in an experience of increased
loudness and associated stress (annoyance). But it would be highly problematic to
postulate a deterministic relationship between measurable environmental factors and
occupants' evaluation of environmental conditions (Mahdavi, 2011a, 1996a, 1996b).
People's evaluation of exposure situations may be easier to describe and
predict in when the material-energetic aspect of the environmental relationships
dominates. In extreme cases of high-intensity exposure, the necessity for protective
regulations is self-evident due to the obvious health hazards for the involved
individuals (e.g., irreversible physical damage to the organism). It is, thus, not
surprising that most efforts toward predicting the outcome of evaluation processes
have focused on the identification of a measurable material-energetic scale (such as
© ASCE
sound pressure level) to which subjective judgments (such as the degree of

annoyance) are expected to correlate. However, the relevance of internal information
processing for the degree of expressed dissatisfaction associated with various
energetic levels of exposure has been demonstrated in experimental psycho-acoustic
experiments (see section 4.3 for a further discussion of this point).
BOUNDARY CONDITIONS
Motivation. We will address the immediate implications of the human factor for the
built environment (i.e., human requirements and behavior) in sections 4 and 5 of this
paper. But it is important that we consider a number of broader questions pertaining
to the environmental implications of social development and human activity
(Mahdavi, 2012): i) How important is the antecedent consideration of population
growth, lifestyle development, as well as agricultural and industrial production for
the effectiveness of sustainable building efforts? ii) How do the relative resource
needs and environmental loads associated with building activity compare to other
domains of human activity and production such as industry and transportation? iii)
To which extent can contextual factors such as urban planning decisions and
mobility solutions affect and constrain the energy and environmental performance of
individual buildings? iv) How should we account for the impact of user behavior
(including the rebound effect) on the energy and environmental performance of
buildings?
Population and life style. United nation's data (medium growth scenario) projecta for
2050 alone for India and China a combined population number of three billions.
Likewise, the population of Africa – slightly over 200 millions around 1950 – is
projected to approach two billion by 2050. The topic of population growth containment
appears thorny and difficult politically. But even if population growth is seen as an
inevitable and unalterable process, at least the implications for resources, environment,
as well as ecological and social systems should be frankly discussed, rather than
evaded. In human ecological parlance, the ecological valency of an ecosystem can
sustainably support only a finite number of people of a given ecological potency.
Transgressing that limit invariably results in an ecological degradation of the
environment.
The ecological strain resulting from population increase is aggravated by a
parallel process involving the improvement of living standards – at least for some
populations – around the world. For instance, the global primary energy consumption
of China is projected to increase – from the value in 2008 – about 230% to reach
roughly 200 exajoules by the year 2035 (EIA, 2012). Moreover, whereas in the last
twenty years per capita energy use has been stagnating (albeit at a very high level) in
countries such as United States and Germany, both per capita energy use and Gross
National Income (GNI) have been increasing in China, India, and Brazil (Databank,
2012). In fact, the Gross National Product (GNP) of China and India is projected to
increase within a period of 40 years (2009 – 2049) roughly by a factor of 6 and 9
respectively. Lest these assertions are misunderstood: Rise in people's standard of
living is necessary, crucial, and desirable socially and ethically. However, even
though not a necessity, per capita improvement in living standard is typically
mirrored in per capita increase in resource depletion and environmental impact. The
environmental ramifications of such developments can be easily exemplified.
Human ecologically speaking, rise in indicators such as GNP and GNI can be
interpreted as a population's increased ecological potency. A precarious implication
© ASCE
of this rise is the population's higher capacity to rapidly exploit natural resources and
intervene in the working of natural systems, resulting in all too familiar negative
consequences. Thus, the combined ecological and environmental effects of
concurrent population growth and the rise of the so-called standard of living can
easily undo incremental efficiency improvements in the building sector.
Buildings, Industry, mobility. The relative share of buildings in energy use and
environmental impact, as compared to other areas human activity and production such
as industry, agriculture, and transportation is undoubtedly significant. In European
Union, the relative energy demand for the domains transportation, industry, and
buildings (residential and commercial sector) was in the year 2000, 31%, 28%, and
41% respectively (Janssen, 2004). Comparing buildings' share in resource depletion
with other areas serves here two purposes: At a global level, the investment costs and
efforts for efficiency improvement potential in each domain can be compared to those
in the other domains. Thus, public funds and financial incentives such as tax incentives
could target those areas where maximum energy efficiency improvement and
environmental impact reduction can be achieved. At a more detailed level, where
behavioral decisions of individuals may matter, building-related energy use and
environmental impact issues could be assessed considering individual energy use
profiles and life styles.
This latter point can be further illustrated if we consider the relative energy
allocation to various activities of middle-class individuals in a European country
(Mahdavi, 2010). It seems activities such as driving cars, travelling with airplanes, and
using electronic gadgets are not all insignificant as compared to energy requirements for
heating of buildings. As mentioned earlier, the so-called developing countries
increasingly adapt both production and consumption practices of the so-called
developed countries. A significant case in point in this regard pertains to the automobile
industry. It is expected that the global demand for automobiles will increase from
currently around 70 millions to around 110 in the year 2020. Thereby, the highest rate
of increase will occur in growing markets such as China. An even more irrational case
of energy and resource usage pertains to the world-wide production and deployment of
weapons. Expenditures for weapons dramatically increased in the years 2001 to 2011
(over 80% in USA and Russia: close to 190% in China, around 60% in Saudi Arabia
and India).
The decisive role of the human factor in the sustainability discourse is not
limited to population explosion and increasing affluence in some populations around
the world. A highly important – but insufficiently understood – variable concerns the
behavior of individuals and populations. Broadly speaking, people's behavioral
tendencies may in certain cases favor solutions and products that are disadvantageous
from the energy and environment point of view. For instance, in various analyses of
human mobility (see, for example, Knoflacher, 1996), people, confronted with multiple
options to embark on a trip, have been shown to display a tendency to favor the
reduction of their personal physical exertion. In other words, people may prefer
mobility options such as driving a car (even if they may require – for instance while
searching for a parking space – more time), rather than walking and biking. On the
other hand, the choice of a certain mobility medium also depends on the availability of
options. If a regional and urban setting provides an extensive infrastructure tailored for
© ASCE
individual transport, it should not come as a surprise if cars would emerge as the
dominant mobility medium. In contrast, increased provision of well-designed and
convenient pedestrian routes and bike paths has shown to result in a correspondingly
higher rate of trips made on foot or on bike. Behavioral patterns with implications for
sustainable development can be influenced, to a certain degree, not only by
consciousness raising measures and information campaigns, but also – and perhaps
more effectively – via proper design and planning measures and strategies as well as
proper economic incentives.
Human behavior and the built environment. The ramifications of people's presence
and actions in buildings are not sufficiently documented. Occupants operate buildings'
control devices such as windows, shades, luminaires, radiators, and fans to bring about
desirable indoor environmental conditions. These control actions can significantly
influence buildings' performance (Mahdavi, 2011a, 2011b). A better understanding of
the patterns of human presence and control-oriented behavior can conceivably facilitate
and guide both technical responses such as occupancy-sensitive environmental control
systems and information campaigns toward improving buildings' energy efficiency and
environmental performance (see chapter 5 for a detailed treatment).
A further example for the relevance of human behavior in the sustainable
buildings discourse concerns the so-called rebound effect (Sorrell, 2007). This effect
refers to the paradoxical circumstance that, under certain conditions, energy
efficiency measures may end up increasing the energy use (see section 5.4).
Instances of rebound effect in the building domain underline the importance of the
human factor in energy efficiency potential assessments and projections. For the
expected efficiency effects of technical measures to materialize, behavioral and
economical boundary conditions must be taken into consideration. This also explains
why it may be a good idea to adjust energy prices in tandem with energy efficiency
improvements.
Summary reflections. The above discussion of the boundary conditions suggests that
there may be an extensive set of essential higher-level measures and actions involving
human populations and individuals that could be undertaken parallel with, if not prior to
measures focused on individual buildings: i) Population increase needs to be addressed
in earnest. Whether the focus is on a region, a country, a continent, or the entire planet,
the respective ecological valency or the available carrying capacity must be considered.
ii) In policy, in education, and all manners of social discourse, the distinction between
the standard (or quality) of living and purely economic measures such as GNP must be
taken seriously. When a society's entire economic and political system is exclusively
focused on narrowly defined monetary gains, it should not come as a surprise if genuine
sustainability concerns can be forgone in favor of superficial and ancillary agenda. iii)
The existing mobility paradigm needs to be reexamined. Individual energy and
emission intensive modes of motorized transportation need to be radically reduced. It is
essential to move away from the practice of designing and organizing cities around cars,
instead of around pedestrians. iv) Individuals and communities, especially in affluent
societies, need to made aware of the environmental consequences of their life style
choices in view of mobility, residency, diet, and recreational activities. v) Overarching
© ASCE
sustainability considerations play an important role also in the narrowly defined

building construction domain. Are sustainability issues considered in the applicable
urban planning prerequisites and land use regulations? Is the intended construction
activity (e.g., erection of specific building) motivated by a genuine necessity? Could
such necessity be accounted for through retrofit and adaptation of existing buildings
instead of new construction?
BUILDING PERFORMANCE AND HUMAN REQUIREMENTS
Introductory remarks. In building science, the habitability of indoor environments is

often treated in distinct domains involving, for instance, thermal, visual, and acoustical
aspects of building performance. This is prudent as a matter of expediency, as long as
we keep in mind that an aggregate judgment of indoor environment quality is the
complex result of material-energetic and information-related processes pertaining to all
such domains. From a human ecological perspective, we can ask ourselves two
interrelated questions, one relevant to people's ecological potency and the other to the
ecological valency of their surrounding outside world: i) Which characteristics of
human beings' are relevant to how they cope with and evaluate (thermal, visual,
acoustical, …) conditions around them? ii) What aspects of people's surrounding indoor
environment are relevant – and should be targeted in the building design and operation
processes – to accommodate people's needs and expectations? To answer these
questions, an understanding of the physics, physiology, and psychology of the processes
involved would be required. To illustrate some of the issues involves, we focus here on
the thermal aspects of indoor environments.
The thermal environment

Heat balance and human body. Under normal conditions, the human thermoregulation
system can maintain a fairly constant core body temperature. Toward this end, the
human organism needs long term energy balance in the course of thermal exchange
with the environment. Specifically, energy losses to the environment need to be
compensated via internal heat production via human metabolism. The metabolic rate
and the associated rate of heat transfer is primarily a function of the activity level and
the surrounding's temperature.
Fluctuations in body's thermal balance with its surroundings occur frequently
and are generally harmless as long as they are not long-term or severe. However, if due
to a sustained imbalance the body's core temperature cannot be maintained, not only
thermal discomfort, but also critical physiological consequences may arise. Multiple
mechanisms are involved in human body thermoregulation. For instance, blood flow
through skin tissue can be modulated by constricting and dilating blood vessels.
Sweating can increase evaporative heat loss from the skin. Heat production can be
increased through shivering, which signifies increased metabolism in muscles. Given
the importance of long term thermal balance between human body and its surrounding,
human thermal sensation has been interpreted as a kind of warning system. Departure
from the thermal equilibrium state would be experienced as thermal discomfort,
motivating people to seek conditions or apply measures toward regaining thermal
equilibrium. Thermoreceptors distributed over the skin provide the brain with
© ASCE
information on skin temperature. In general, thermoreceptors of the human skin give

rise to the conscious impressions of environments as being warm or cold. Thereby, cold
and warm sensations may be perceived as positive or negative depending on the body
core temperature. If the core is overheated, a cold sensation is likely to be perceived as
pleasant. However, if the core is cold, a warm sensation is more likely to be
experienced as pleasant.
Thermal comfort models and standards. Insights gained based on the study of the
human body's thermoregulation have informed the attempts to identify relationships
between physiological factors and the experience of thermal comfort. For example,
indices such as the PMV (Predicted Mean Vote) and PPD (Predicted Percentage of
Dissatisfied) postulate that people would be likely to be in the state of thermal
comfort if their mean skin temperature and sweat secretion are within a certain range
(Fanger, 1972). PMV is a numerical index that expresses the statistical mean of
thermal comfort evaluations by a large group of people. The correlating PPD index is
a predictive measure of the percentage of thermally dissatisfied people in a specific
thermal environment. Based on studies conducted in climate chambers, optimal
thermal comfort conditions were suggested to correlate with personal factors
(metabolic rate, clothing) and environmental conditions (air and radiant
temperatures, humidity, air flow speed). The results of such studies have been
frequently structured in terms of equations and associated rules, tables, and graphic
means that can be used by designers and engineers to infer preferable indoor climate
conditions for people with a certain level of clothing and activity. Various standards
such as ISO 7730 (ISO 2005) and ASHRAE 55 (ASHRAE, 2004) specify building
categories based on PMV ranges. The PMV-based approach is meant to statistically
apply to large groups of people. Moreover it primarily pertains to the human body as
a whole. Further studies and associated formulas have addressed the local thermal
discomfort due to draught risk, radiative asymmetry, cold or warm floor surfaces,
and vertical temperature gradients (Fanger et al., 1988, 1985; Olesen et al., 1979;
Olesen, 2008, 2002).
The above approach to description and prediction of thermal comfort has both
merits and limitations. Amongst the virtues of the approach is the systematic way in
which in a number of personal (level of activity, thermal resistance of the clothing)
and environmental factors (air temperature and humidity, air flow speed, radiant
temperature) serve as the input information to predict thermal comfort level of
people in a room. However, the approach has also been faulted with regard to
intrinsic limitations and predictive performance. On the one hand, precise definition
of the model's input variables (e.g., the exact determination of people's activity or the
thermal resistance of their clothing) is not trivial and may constrain the applicability
of the model to practical situations. On the other hand, laboratory-based models may
not properly capture conditions in the field, where inhabitants are typically adapted
to their living and working environments. Moreover, the steady-state assumptions
underlying classical thermal comfort models do not apply to real – particularly free-
running – buildings. Field studies have documented, particularly in free-running
buildings, considerable deviations of PMV-based thermal comfort predictions from
actual comfort votes by the occupants (see, for example, de Dear and Brager, 2002).
In this context, human-ecologically relevant psychological factors such as people's
expectations as well as their behavioral adaptations have been suggested to play a
key explanatory role. People, when thermally uncomfortable, tend to react in such a
way as to reduce thermal constraint and thus restore thermal comfort (Humphreys
and Nicol, 1998; Auliciems, 1983). Such reflections have led to the conception of an
adaptive approach to thermal comfort definition and prediction (Nicol et al., 2012).
© ASCE
Numerous actions may be initiated – consciously or unconsciously – by buildings'

occupants as response to cold and heat sensations. For example, people can modify
body heat generation (e.g., by reducing or increasing the activity level), control body
heat loss (e.g., by changing posture or changing the clothing level), regulate the
thermal conditions in the environment (e.g., by changing the thermostat settings, or
operating windows, fans, and shades), and change location in the space or building.
Field surveys and field studies appear not only to point to the limitations of simple
heat-balance based comfort models, but also reveal certain adaptive relationships.
Specifically, neutral (or comfort) temperature expressions collected in various
surveys display a significant relationship to the prevailing operative temperatures.
This has been interpreted to imply that people can generally match their comfort
temperature to the conditions in their environments (Nicol et al., 2012). Moreover,
field survey results – particularly from buildings in free-running mode – also point to
a strong correlation between neutral temperatures and prevailing outdoor
temperatures (Humphreys at al., 2010).
Considerations pertaining to the aforementioned adaptive processes have
found their way into comfort guidelines. Thereby, acceptable indoor temperatures in
free-running buildings are defined as a function of the outdoor air temperature.
Thereby, various methods have been proposed to derive running averages for
outdoor temperature. For example, EN ISO 15251 (EN ISO, 2007) includes three
categories of thermal comfort in naturally ventilated buildings as a function of the
running average of – exponentially weighted – daily outdoor temperature.
Views on applicable thermal comfort zones have major consequences for
sustainability considerations in buildings. If the assumptions regarding indoor
environmental conditions that would result in acceptable thermal conditions in rooms
are too rigid, free-running buildings would not be a viable option. Generally
speaking, only buildings with HVAC systems can provide – at least in principle –
narrowly controlled indoor environmental conditions independent of the outdoor
conditions. However, both examples from vernacular architecture (Mahdavi, 1996c)
and more recent low-energy buildings have shown, that in many climatic zones and
over considerable periods of time, properly designed buildings can offer adequate
indoor conditions without the need for extensive mechanical systems for heating,
cooling, and ventilation. Surveys and field study results underlying the adaptive
approach to thermal comfort postulate a more flexible framework for the definition
of desirable indoor environmental conditions in buildings. They help thus counter
comfort-based (and more recently, productivity-based) arguments for deployment of
large-scale energy-intensive mechanical environmental control systems in buildings.
A note on the information-related aspect of environmental relationships. While the

efforts to ground the adaptive approach to thermal comfort on a solid scientific basis
(Nicol et al., 2012) are not entirely convincing, insights from associated studies are
valuable. They suggest that multi-level complexities involve in human processes of
environmental sensation, perception, and evaluation cannot be captured with simple
heat-balance based models of thermal comfort. Future advances could shed more light
on the combined physiological, psychological, and cognitive underpinnings of thermal
comfort evaluation processes. Meanwhile, however, building design and operation
professionals must be sensitive to the considerable variance in thermal responses of
people to similar thermal conditions. Human ecology provides a conceptual high-level
(and qualitative) perspective toward the understanding of such evaluative variances in
terms of the importance of not only the matter-energetic aspect but also the information-
related aspect of the environmental perception phenomena.
© ASCE
As mentioned earlier, one should be utterly careful with postulating strictly

causal relationships between measurable environmental factors (such as temperature
and humidity in a room) and occupants' evaluation of environmental conditions (i.e.,
the thermal comfort vote). A number of studies have aptly demonstrated the
important effects of information-related aspects of environmental exposure situations
on perception and evaluation processes (Mahdavi, 2011a). For example, in one
experiment, two demographically similar groups of participants provided
significantly different assessments of the same acoustical event (recorded white
noise). Participants in the first group, who were told the recording was of a waterfall,
judged it much more favorably than the second group, who was told the recording
was of a factory. People's attitude toward the alleged source of an acoustical event
clearly influenced their evaluation of the exposure, despite the absence of any
objective difference in the nature of the event (Mahdavi, 2011a).
In another experiment (Schönpflug, 1981), participants were exposed to white
noise of different intensity while performing certain tasks (time estimations). The
participants who received positive feedback about their performance ranked the same
acoustical exposure as less annoying than those who received negative feedback
concerning their performance. But the feedback messages were manipulated and did
not reflect the true performance. Hence, their effect on participants' subjective
evaluation of the noise exposure situation cannot be explained in terms of an
acoustically induced impairment. The explanation lies rather in the nature of the
information processing that was triggered by the combined effect of acoustical
exposure and negative feedback. The degree of annoyance due to noise was
apparently higher, once it was identified as the reason for one's (alleged) failure.
Such experiments imply that subjective evaluations are not at all fully determined by
energetic descriptors of the so-called environmental exposure. Rather, such
evaluations emerge through the complex workings of the information processing in
human minds.
BUILDING PERFORMANCE AND HUMAN IMPACT
The relevance of people's behavior. To appreciate the critical role of the people factor
for building performance, consider a few simple questions (Mahdavi, 2011b) that
building designers are typically expected to answer: i) How much energy will be
needed to heat, cool, ventilate, and illuminate buildings? ii) What kinds of indoor
conditions concerning thermal comfort and air quality are to be maintained in
buildings? iii) What level of daylight can be expected in indoor environment under
dynamically changing outdoor illuminance levels? iv) Will the acoustical environment
in indoor spaces provide the necessary conditions for communication and task
performance? v) Can occupants be safely evacuated from buildings in case of an
emergency such as outbreak of fire?
Obviously, none of these questions can be reliably answered without
considering the role of the people living and working in buildings. People affect the
performance of buildings, due to their presence and their actions. Energy and thermal
performance of buildings is not only influenced by the people's presence as a source
of sensible and latent heat, but also due to their actions, including use of water,
operation of appliances, and manipulation of building control devices for heating,
cooling, ventilation, and lighting. User-based operation of luminaires and shading
devices in a room affect the resultant solar gains, light levels, and visual comfort
conditions. Presence of people in a room and the associated sound absorption
influences the acoustical performance of the room. Safety performance of a building
© ASCE
cannot be evaluated without considering the behavior of people under emergency.

People's relevance to performance-based design is, however, not restricted to
assumptions regarding user presence and actions in buildings. Frequently,
information on user-based requirements in view of thermal, visual, and acoustical
comfort must be explicitly reflected in building operation specifications. For
example, control settings for air temperatures and illuminance levels in architectural
spaces must be defined in accordance with knowledge on people's needs and
requirements. Moreover, targeted values of performance variables are expected to
encapsulate information that is relevant to people's requirements and expectations in
view of health, comfort, and satisfaction. Appropriate selection and interpretation of
building performance indicators requires thus that the relationship between occupied
spaces and their occupants are considered and understood. Given this background,
we need to systematically situate buildings' users and occupants in the context of
performance requirements. Accordingly, we must deal with the mechanisms and
corresponding models of how people's presence and interactions with buildings'
environmental systems influence the values of relevant performance indicators.
Two kinds of effects. Broadly speaking, a useful distinction can be made between
"passive" and "active" effects of users and occupants on buildings' performance.
Passive effects of people on indoor conditions in buildings denote those effects caused
by the mere presence of people in the building. For instance, hygro-thermal conditions
and indoor air quality in architectural spaces are influenced by such passive people
effects: Depending on their activity, people release not only various quantities of
sensible and latent heat, but also water vapor, carbon dioxide, and other execrations and
odorous substances. Likewise, in the building and room acoustics domain, presence of
people in a space has an effect on the sound field via introduction of additional sound
absorption. To capture the passive effects of people's presence in buildings in the design
process, we typically rely on existing data such as occupancy load schedules derived
from measurement results of people's metabolic rates. This is as such a straight-forward
process, barring two possible complexities. Firstly, different levels of resolution are
conceivable regarding temporal and spatial distribution of such passive effects.
Secondly, the passive people effects such as heat emission may depend on the context,
e.g., thermal conditions in occupants' rooms. This interdependence would require the
concurrent consideration of the human agent and its immediate environment.
In most buildings, occupants operate control devices such as windows, shades,
luminaires, radiators, and fans to bring about desirable indoor environmental conditions.
We refer to these control actions as people's active effects. They have a significant
impact on buildings' hygro-thermal and visual performance. To predict and evaluate
buildings' performance we need good knowledge of such control-oriented user
behavior. General information about building type (residential, commercial) and
environmental systems (free-running, air-conditioned) as well as organizational and
administrative information (e.g., working hours) can only provide rough directions
regarding such active effects. More representative people presence and action models
require, however, extensive observational data based on empirical studies of occupancy
and control-oriented user behavior in a large number of buildings. Thereby, possible
relationships between control actions and environmental conditions inside and outside
buildings could provide the underlying basis for derivation of user behavior models.
© ASCE
Empirical observations and models. There is a substantial and growing body of

observational studies to capture the patterns of occupants' presence in buildings and
their interactions with buildings' environmental control systems such as windows,
blinds, and luminaires. Frequently, such studies attempt to establish a link between user
control actions or the state of user-controlled devices and measurable indoor or outdoor
environmental parameter (see, for example, Hunt, 1979; Love, 1998; Reinhart, 2004;
Boyce, 1980; Lindelöf and Morel, 2006; Rea, 1984; Inoue et al., 1988; Herkel et al.,
2005; Nicol, 2001; Mahdavi, 2011b). While highly useful, these studies are often
variously limited, due – amongst other things – to the small number of buildings and
rooms involved, the duration and consistency of data collection, the accuracy of the
measurements, the robustness of the analyses, and the clarity of the documentations.
Some of these limitations and their implications were addressed in the course of a
recent case study, involving a number of office buildings in Austria. Thereby, we
systematically collected an extensive set of observational data regarding building
occupants' presence and control action patterns pertaining to lighting and shading
systems while considering the indoor and outdoor environmental conditions under
which those actions occurred (Mahdavi et al., 2008a, 2008b; Mahdavi, 2011b). Some of
the lessons learned from this study are summarized below.
It is important to understand that the pattern of people's presence in buildings
cannot be simply inferred from building type and function – e.g. residential versus
commercial. Nor can it be based solely upon organizational information from
building and facility managers. In our study of five office buildings in Austria, the
mean occupancy patterns was unlike common assumptions in the professional
community or presumptions of the organizations involved. Moreover, the buildings
we studied displayed very different occupancy patterns. Even if all offices in a
building belong to the same organization, there could be drastic differences between
their occupancy patterns. A building's usage and the functions it supports can
repeatedly and considerably change over time, yet again implying variable and
hardly predictable occupancy patterns. Moreover, offices can be, in the course of
time, assigned to different individuals or user groups with inherently different
occupancy tendencies. Ultimately, the same individual occupant might, over time,
display varying patterns of presence, given professional or personal circumstances.
Such factors lead to the considerable uncertainty in assumptions pertaining to
occupancy levels to be expected in buildings.
Our specific case study did result in a number of empirically-based
statistically significant relationships between the frequency or probability of user
control actions (involving, for example, lights, blinds, windows) on the one hand and
some independent variables pertaining to occupancy, indoor environment, or outdoor
conditions on the other hand (Mahdavi 2011b). Such empirically-based models might
provide clues and indications concerning the environmental triggers of behavioral
tendencies. But they certainly do not represent causal models of human control
actions in buildings. Moreover, as with all statistically derived relationships, these
kinds of models are limited in at least two regards:
First, they cannot be divorced from the population from which they are
derived and simply applied to other contexts, at least not without losing much of their
statistical credence. Behavioral tendencies and their dependencies on hypothesized
independent variables are influenced by a large number of diverse factors, such as
the climate, cultural issues, building type and functions, organizational specifics,
building systems peculiarities, space orientation, and interior design features
influence. Second, aggregate models do not explicitly reflect the inherently
© ASCE
probabilistic nature of most control-oriented control actions. Nor do they capture the
dynamism of actual processes and events in buildings, as stochastic models can – at
least in principle (Fritsch et al., 1990; Nicol, 2001).
The latter models have been used to generate time series of both occupancy
intervals and user control actions that "look" similar to actual (real) processes and
event sequences. Thus, if grounded in quantitatively sufficient and qualitatively
adequate empirical data, stochastic occupancy and control action models could
capture the general occupancy-triggered processes in a building. Such models can be
implemented in simulation applications in terms of autonomous agents with built-in
methods to generate behavioral patterns that appear realistic (Bourgeois, 2005;
Chang and Mahdavi, 2002; Liao et al., 2011). However, the promise of stochastic
occupancy needs to be qualified against both reliability and applicability concerns
(Mahdavi and Tahmasebi 2015).
An argument can be made for the utility of simple (code-base or descriptive)
occupancy-related simulation input assumptions in the design development phase,
where calculations can be used to obtain numeric values for a number of aggregate
performance indicators such as buildings' annual heating and cooling loads. Such
aggregate indicators can support at least two purposes: i) benchmark a specific
building design proposal against applicable codes, standards, and guidelines, or ii)
comparatively assess the likely performance of multiple design alternatives. Thereby,
concise inferences are expected concerning the quality of the proposed building
"hardware" vis-à-vis design variables pertaining to the building's envelope, massing,
orientation, shape, construction, etc. Naturally, this is done under "standardized"
conditions pertaining to external climate, which is typically represented in terms of a
standard weather file, and internal occupancy-related processes, which are typically
represented in terms of fixed, more or less detailed assumptions regarding internal
gains, ventilation rates, etc. Theoretically speaking, the use of a probabilistic
presence and user action models could represent a problem not only for code-based
compliance checking, but also for the performance analyses of design alternatives,
when the aim is to compare multiple alternative designs irrespective of variance in
contextual boundary conditions (weather) and occupancy.
A different circumstance arises, however, if we consider more elaborate
building design analysis scenarios, which require us to consider the implications of
uncertainties associated with occupancy processes in buildings. Differences in
occupancy patterns over time and location can be quite significant. Such differences
can be important especially in view of the variance of thermal loads or conditions in
various zones of a building. Information regarding temporal and zonal load
variations is critically important, for example when calculations provide essential
data for design and sizing of indoor climate control systems. Thus, rigid models of
user presence and behavior that ignore associated stochastic fluctuations and the
resulting uncertainties would be rather problematic, if the detailed configuration of a
building's mechanical equipment is the main concern: While dealing with the
requirement of providing sufficient heating and cooling capacity to different zones of
a building, the variability of required thermal loads cannot be captures via spatially
and temporally averaged occupancy assumptions.
It seems as though different approaches to representation of occupancy-
related processes in building design support may be appropriate given different
scenarios. If consideration of the implications of variance in model input
assumptions is evidently critical to a specific performance inquiry, then probabilistic
models of occupancy presence and control actions would be appropriate. On the
other hand, when the objective of a performance analysis inquiry is to benchmark
design proposals against applicable codes and standards or to parametrically compare
design alternatives, uncritical inclusion of random variations of boundary conditions
© ASCE
and internal processes in inquiries may be unnecessary if not counterproductive.
A note on the rebound effect. As noted earlier (see section 3.4), studies in different
fields have shown that energy efficiency improvement measures may miss their targets,
sometimes considerably. Paradoxically, energy efficiency measures may even end up
increasing the energy use. This circumstance is often referred to as the "rebound effect".
It can be defined as the non-materialized fraction of the projected energy saving due to
an energy efficiency improvement measure. Related observations have been also
reported in the building domain, whereby projected energy performance of new
buildings and the energy saving potential of thermal retrofit measures on existing
buildings were found to be overly optimistic. The rebound effect is complex and can
have multiple roots. Thereby, one of the contributing factors may be attributed to
behaviorally relevant circumstances. For example, thermal retrofit measures on a
building can principally reduce the heating energy required to maintain certain thermal
conditions in that building (Housez et al. 2014). However, the reduction potential may
not be exploited in actuality, if occupants modify their behavior in a more energy-
intensive direction. For example, they may change the temperature settings for heating,
or they may ventilate spaces more frequently, or they may turn on the heating in more
rooms. Such behavioral phenomena may explain the results of a number of recent
studies, which documented a lower than expected energy efficiency improvement effect
following thermal retrofit measures pertaining to existing buildings. The rebound effect
may also involve the redirection of energy efficiency gains in one area to increased
consumption in another. For example, monetary benefits from building-related energy
conservation measures in terms of lower heating costs may be redirected toward
increased energy use in higher fuel usage for car driving.
Summary remarks. The importance of people's passive and active effects on building
performance (e.g., indoor environmental conditions, energy use) can be significant.
Accordingly, many recent and ongoing research efforts attempt to understand and
predict passive and active occupancy effects on building performance. Thereby,
physiological and psychological descriptions of occupancy as well as empirically-based
observational data provide the knowledge base. Specifically, long-term high-resolution
empirical data on people's presence and control-oriented actions in buildings can
support the generation of general patterns of user control behavior as a function of
indoor and outdoor environmental parameters such as temperature, air flow, air quality,
illuminance, and irradiance. These patterns can be expressed either in terms of
typologically differentiated aggregate occupancy and control action models or realized
in terms of emergent behavior of a society of computational agents with embedded
stochastic features. Future developments in this area are expected to facilitate detailed
computational models of environmental processes in buildings via comprehensive
multiple-coupled representations that dynamically capture the states of occupancy,
building, and context.
CONCLUSION
In this paper, we reviewed the role of people in the built environment from a
number of different vantage points. Thereby, we explored the potential of human
© ASCE
ecology as a conceptual framework for our inquiry. In this framework, building

activity may be interpreted as an instance of largely regulatory activities by human
beings to better match the ecological valency of their immediate surroundings with
their ecological potency. Moreover, we suggested that if the notion of sustainable
architecture is to be taken seriously, it must critically address not only the technical
quality of individual buildings and the requirements and behavioral patterns of their
inhabitants, but also the broad ramifications of population growth, life style
development, climate change, social policies, investment priorities, and urban
planning.
Given this multi-faceted treatment, it is fitting to conclude a paper dedicated
to the role of people in the built environment with a thought concerning a specific
group of people, namely those involved in research, education, development,
production, and management activities pertaining to building delivery and operation
processes. This group includes, amongst others, scientists, trainers, developers,
architects, engineers, and facility managers. Members of this group should of course
make every attempt to target habitability and sustainability of the built environment
via knowledge-based design, ecologically informed selection of building materials
and components, as well as in-depth consideration of occupants' requirements. But
viewed more broadly, we need to remind ourselves that the professionals in the
domain of built environment do not operate in a vacuum. Rather, they are
constrained by a large number of economical, procedural, and political boundary
conditions. In many instances, the role of professionals in building projects starts
only after a number of crucial decisions – with decisive sustainability ramifications –
have been made. If concrete building projects start at the point where it is already
decided which new buildings have to be erected, where they should be located, and
under which urban setting, then the degrees of freedom for building designers and
engineers to contribute meaningfully to sustainability efforts are already severely
limited. If responsible building professionals are consulted and involved only late in
the consequential chain of environmentally relevant decisions in the building sector,
their role would be reduced to "damage control". But as responsible citizens, those
involved in the building delivery and operation process need to enter the socially
relevant sustainability discussion at the earliest opportunity, where processes are
initiated and decisions are made that fundamentally influence the habitability and
sustainability of the built environment.
REFERENCES
ASHRAE Standard 55 (2004). “Thermal environmental conditions for human
occupancy”. American Society of Heating, Refrigerating and Air-
Conditioning Engineers.
Auliciems, A.(1983) “Psychological criteria for global thermal zones of building
design.” International Journal of Biometeorology. 27 pp.69 – 86.
Bourgeois, D. (2005) “Detailed occupancy prediction, occupancy-sensing control
and advanced behavioral modeling within whole-building energy
simulation.” PhD Thesis, Université Laval, Quebec, Canada.
Boyce, P.(1980) “Observations of the manual switching of lighting.” Lighting
Research & Technology, 12:4 pp.195-205.
Chang, S. and Mahdavi, A.(2002) “A hybrid system for daylight responsive lighting
control.” Journal of the Illuminating Engineering Society, 31:1 pp.147- 157.
Databank World databank.(2012)
URL: http://databank.worldbank.org/ddp/home.do?Step= 1&id=4, (accessed
February.
de Dear, R.J. and Brager, G.S.(2002) “Thermal comfort in naturally ventilated
© ASCE
buildings.” Revisions to ASHRAE Standard 55, Energy and Buildings, 34:6

pp.549 – 561.
EIA (2011) “International energy outlook.”
URL: http://www.eia.gov/forecasts/ieo/ ieo_tables.cfm.
EN ISO 15251 (2007) “Indoor environmental input parameters for design and
assessment of energy performance of buildings addressing indoor air quality,
thermal environment, lighting and acoustics.”
Fanger, P.O. (1972) “Thermal Comfort.” McGraw-Hill Book Company, ISBN 0-07-
019915-9.
Fanger, P.O., Melikov, A.K., Hanzawa, H., Ring, J. (1988) “Air turbulence and
sensation of draught.” Energy and Buildings, 12:1, pp.21 – 39.
Fanger, P.O., Ipsen, B.N., Langkilde, G., Olesen, B.W., Christensen, N.K., Tanabe,
S. (1985) “Comfort limits for asymmetric thermal radiation.” Energy and
Buildings 8, (1985) pp.225 – 236.
Fritsch, R., Kohler, A., Nygard-Ferguson, M., Scartezzini, J. L. (1990) “A stochastic
model of user behaviour regarding ventilation.” Building and Environment,
25:2 (1990): 173-181.
Gibson, J. (1979) “The Ecological Approach to Visual Perception”, ISBN 0-89859-
959-8.
Gibson, J. (1977) “The Theory of Affordances.” In Perceiving, Acting, and Knowing,
Eds. Robert Shaw and John Bransford, ISBN 0-470-99014-7.
Herkel, S., Knapp, U., Pfafferott, J. (2005) “A preliminary model of user behavior
regarding the manual control of windows in office buildings.” Proceedings of
The Ninth International IBPSA Conference, Building Simulation, Montréal,
Canada, pp. 403-410.
Housez, P., Pont, U., Mahdavi, A. (2014) “A comparison of projected and actual
energy performance of buildings after thermal retrofit measures.”;
Journal of BUILDING PHYSICS, 38 (2014), 2; S. 138 - 155.
Humphreys, M.A. and Nicol, J.F. (1998) “Understanding the adaptive approach to
thermal comfort.” ASHRAE Transactions, 104:1 pp.991 – 1004.
Humphreys, M.A., Rijal, H.B., Nicol, J.F. (2010) “Examining and developing the
adaptive relation between climate and thermal comfort indoors.” Proceedings
of Conference on Adapting to Change: New Thinking on Comfort, Windsor,
UK, (http://nceub.org.uk).
Hunt, D.(1979) “The use of artificial lighting in relation to daylight levels and
occupancy.” Building and Environment, 14, pp.21-33.
Inoue, T., Kawase, T., Ibamoto, T., Takakusa, S., Matsuo, Y. (1988) “The
development of an optimal control system for window shading devices based
on investigations in office buildings.” ASHRAE Transaction, 94, pp.1034-
1049.
ISO 7730 (2005) “Ergonomics of the thermal environment – Analytical
determination and interpretation of thermal comfort using calculation of the
PMV and PPD indices and local thermal comfort criteria.”
Janssen, R. (1996) “Towards energy efficient buildings in Europe.” (Final Report).
EuroACE (the European alliance of companies for energy efficiency in
buildings).
Knoflacher, H. (1996) „Zur Harmonie von Stadt und Verkehr: Freiheit vom Zwang
zum Autofahren.“ Böhlau. ISBN 3-205-98586-9.
Knötig, H. (1992a) “Human Ecology - The exact science of the interrelationships
between Homo sapiens and the outside world surrounding this living and
thinking being.” The sixth meeting of the Society for Human Ecology "Human
Ecology: Crossing Boundaries". Snowbird, Utah, USA.
Knötig, H. (1992b) “Some essentials of the Vienna School of Human Ecology.”
© ASCE
Proceedings of the 1992 Birmingham Symposium; Austrian and British

Efforts in Human Ecology, Archivum Oecologiae Hominis, Vienna, Austria, .
Liao C, Lin Y, Barooah P. (211) “Agent-based and graphical modelling of building
occupancy.” Journal of Building Performance Simulation, doi:
10.1080/1940493.2010.531143.
Lindelöf, D. and Morel, N. (2006) “A field investigation of the intermediate light
switching by users”, Energy and Buildings, 38(2006) pp.790-801.
Love, J.A. (1998) “Manual switching patterns observed in private offices.” Lighting
Research & Technology, 30:1, pp.45-50.
Mahdavi, A. (2012) “Sustainable buildings: some inconvenient observations.”
Keynote in Proceedings First International Conference on Architecture and
Urban Design; 1-ICAUD", EPOKA Univ.; Dept. of Arch. (ed.); Epoka
University Press, 1, ISBN: 9789928-135-01-8.
Mahdavi, A. (2011a) “The human dimension of building performance simulation.”
Keynote: Building Simulation2011 - IBPSA 2011, Sydney, Australia;
(Soebarto V, Bennetts H, Bannister P, Thomas PC, Leach D.: Editors). ISBN:
978-0-646-56510-1 (2011a) pp.K16 - K33.
Mahdavi, A. (2011b) “People in building performance simulation.” Building
performance simulation for design and operation (Hensen JLM and Lamberts
R, editors), Spon Press. ISBN13: 978-0-415-47414-6.
Mahdavi, A. (2010) “Was kann das Plusenergiehaus - Eine kritische Betrachtung
über das Plusenergiehaus.“ Baumagazin. 5 (2010), pp.4 - 7.
Mahdavi, A.: (1996a) “Approaches to Noise Control: A Human Ecological
Perspective.” Proceedings of the NOISE-CON 96 (The 1996 National
Conference on Noise Control Engineering), Bellevue, WA, USA. pp. 649 –
654.
Mahdavi, A. (1996b) “Human Ecological Reflections on the Architecture of the
´Well-tempered Environment´.” In Proceedings of the 1996 International
Symposium of CIB W67 (Energy and Mass Flows in the Life Cycle of
Buildings), Vienna, Austria, pp. 11 - 22.
Mahdavi, A. (1996c) “A Human Ecological View of ´Traditional´ Architecture.”
Human Ecology Review (HER), Volume 3, Number 1, pp. 108 - 114.
Mahdavi, A. (1992) “Acoustical Aspects of the Urban Environment.” Aris; Journal
of the Carnegie Mellon Department of Architecture. Volume 1, pp. 42 - 57.
Mahdavi, A. (1989) “Traditionelle Bauweisen in wissenschaftlicher Sicht.”
Bauforum, Vol. 132 pp. 34 - 40.
Mahdavi, A., Tahmasebi, F. (2015): “Predicting people's presence in buildings: An
empirically based model performance analysis.”; Energy and
Buildings, 86 (2015), S. 349 - 355.
Mahdavi, A., Mohammadi, A., Kabir, E., Lambeva, L. (2008a) “Occupants'
operation of lighting and shading systems in office buildings.” Journal of
Building Performance Simulation, 1:1 : 57-65.
Mahdavi, A., Kabir, E., Mohammadi, A., Pröglhöf, C. (2008b) “User-based window
operation in an office building.” In: Olesen, B., Strom-Tejsen, P., Wargocki,
P. (eds) Proceedings of Indoor Air 2008 - The 11th International Conference
on indoor Air Quality and Climate, Copenhagen, paper 177.
Nicol, J. F. (2001) “Characterising occupant behavior in buildings: Towards a
stochastic model of occupant use of windows, lights, blinds, heaters, and
fans.” In Proceedings of The Seventh International IBPSA Conference,
Building Simulation, Rio de Janeiro, Brazil, pp.1073-1078.
Nicol, F., Humphreys, M., Roaf, S. (2012 “Adaptive thermal comfort. Routledge.”
ISBN: 978-0-415-69159-8.
Olesen, B.W.(2008) “Radiant floor cooling systems.” ASHRAE Journal 50:9, pp.16 –
© ASCE
22.
Olesen, B.W. (2002) “Radiant floor heating in theory and practice.” ASHRAE
Journal 44:7, pp.19 – 24.
Olesen, B.W., Schöler, M., Fanger, P.O. (1979) “Discomfort caused by vertical air
temperature.” in Fanger, P.O. and Valbjorn, O. (eds), Indoor Climate, Danish
Building Research Institute, Copenhagen, pp. 561 – 579.
Rea, M. S. (1984) “Window blind occlusion: a pilot study” Building and
Environment, 19:2, pp.133-137.
Reinhart, C. (2004) LIGHTSWITCH-2002: “A Model for Manual Control of Electric
Lighting and Blinds”, Solar Energy, 77, pp.15-28.
Schönpflug, W. (1981) „Acht Gründe für die Lästigkeit von Schallen und die
Lautheitsregel.“ From Akustik zwischen Physik und Psychologie. In
SCHICK, A. (Ed.), Akustik zwischen Physik und Psychologie. Stuttgart
(Klett-Cotta).
Sorrel, S. (2007) “The rebound effect: an assessment of the evidence for economy-
wide energy savings from improved energy efficiency.” UK Energy Research
Centre, Sussex Energy Group for Technology and Policy Assessment.
Uexküll, J. (1920) “Kompositionslehre der Natur.” (Edited by Thure von Uexküll).
Frankfurt am Main.
© ASCE
Occupant Behaviors and Energy Use:

Creating High-Performance People for High-Performance Buildings
Julia K. Day, Ph.D.1
1
Environment, Behavior, and the Built Environment Laboratory (EBBEL),
Department of Apparel, Textiles, and Interior Design, Kansas State University, 225
Justin Hall, Manhattan, KS 66506. E-mail: juliakday@ksu.edu
Abstract
Many high-performance building strategies, such as daylighting and natural

ventilation, offer a wealth of potential benefits to both occupants and owners (i.e.,
increased productivity, satisfaction, cost savings, etc.). However, these benefits may
be compromised by patterns of occupation, especially if the end-users do not
understand how to interact with buildings as intended. Furthermore, social influences
and resulting occupant behaviors may further complicate energy efficiency goals.
A sequential mixed-methods study was conducted to better understand
occupant behaviors, environmental satisfaction, and learning in high-performance
buildings. Results indicated a significant difference between groups; individuals who
had reported effective training were significantly more likely to be satisfied with their
office environment (Day and Gunderson, 2014). Additional qualitative findings offer
further insight into occupant behaviors and interactions. The primary objective of this
paper is to identify motivators and characteristics of occupant behaviors, as they
relate to building energy use.
INTRODUCTION & BACKGROUND
Many high-performance building strategies offer a wealth of potential benefits

to both users and owners (i.e. increased health, productivity, and satisfaction).
However, these benefits are often compromised by patterns of occupation, especially
if the end-users do not understand how to operate high-performance building
controls. Furthermore, energy efficiency goals can be jeopardized by uninformed
occupant behaviors and interactions with the building. For example, building
occupants have frequently reported using personal space heaters at work. A small
space heater uses about 1000-1500 watts per hour, depending on its efficiency and
size (Otter Tail Power Company, 2009).
× ℎ ÷ 1000 × ℎ ( ℎ)
=
So, let us imagine that one person brought a space heater to their office in
Boston, MA that ran at 1,500 watts, at the rate of 9.448¢ / kWh -- then the cost would
be roughly $1.42 per day, if the heater was run for ten hours per day (1,500 x 10 ÷
1,000 x $0.09448). This number does not seem to be significant. However, what if
© ASCE
this one occupant ran the space heater every day during the colder months (October
through February)? This would equate to $213 ($1.42 x 30 days x 5 months).
Perhaps an additional $213 for an electric bill is still not significant for a large office
building. But, what if 10 people operated personal space heaters for five months of
the year ($2,130)? What about 50 people ($10,650)? And, what if ten of those 50
people forgot to turn the space heater off every day when they went home ($13,622)?
The numbers can quickly add up to a more significant total for a company.
In addition, there are other less obvious factors that may quickly increase
energy costs. A temperature sensor, used to control the heating and cooling in a
particular zone of the building, may be placed in the area where the occupant is
operating the space heater. This may throw off the temperature readings in the overall
space, therefore making other occupants cold. Not only can this lead to thermal
discomfort to other occupants, but it may contribute to even more energy use from
the building’s HVAC system as facility managers try to please occupants by turning
up the heating setpoints even higher to compensate. Thermal comfort in buildings is a
complex issue because people have different thermal preferences, and furthermore,
individual behaviors can impact others’ comfort and energy use outcomes.
At first glance, the example above seems simple: someone brought in a space
heater because they were cold. However, in a high-performance building with finely
tuned HVAC setpoints and aggressive energy reduction goals, these behaviors can be
detrimental. The logic behind the example illustrated above can also be applied to
many different types of high-performance strategies in a building such as natural
ventilation (a window left open in the winter leading to extraneous heating), or
daylighting (window blinds left open on the south side of the building on a summer
weekend, which may lead to excess heat gain); both examples lead to superfluous
building energy use that could have been mitigated through informed occupant
behaviors.
Ultimately, in a high-performance building, occupant behaviors and interactions
with the building can negatively or positively affect energy outcomes for a variety of
reasons. The examples above are overly simplified to illustrate a point, however,
these types of occupant interactions within buildings can become more difficult to
predict depending on the types of building strategies / controls, climate, access to
control, social dynamics, cultural factors, comfort, and personal preferences (Brager,
& de Dear, 2000; Cole & Brown, 2009; Goodwin, 2013; Guerin et al., 2011; Hanqing
et al., 2006; Humphreys, 2005; Luo et al. 2014). The goal of this paper is to unpack
some of these complicated issues and to identify important characteristics of occupant
behaviors, as they relate to building energy use, through the lens of a large mixed-
methods study. Specifically, this paper will focus on two of the research questions
from the study that were geared toward occupant behaviors: (1) Why do occupants
interact with the blinds, electric lighting or thermal controls? (2) For what reasons
do occupants choose not to interact with high performance building features?
METHODOLGY
This study followed a sequential explanatory mixed-methods design. The

methodology for the first two phases are reported in great detail elsewhere (Day &
© ASCE
Gunderson, 2014). In short, the first phase of the study included interviews with
experts in the field, literature review, and examination of building databases, the
second phase of the study included a large-scale survey, and the third phase of the
study was comprised of semi-structured interviews and detailed examination of the
building energy use data, photographs, architectural drawings, and other building
specific data.
This paper will primarily focus on the rich qualitative data analyses within the
third phase. The semi-structured and open-ended interview responses illustrated why
and how occupants were behaving or acting in a certain way within their given
buildings. While each building was unique, collecting data across varying types of
climates and high performance buildings helped to strengthen the overall
generalizability of the qualitative results. Furthermore, these results, coupled with the
quantitative data from the second phase, helped to reinforce the overall research
design and validity (Creswell, 2009).
Participants. The participants for the qualitative phase of the study included
individuals that had completed the survey in the first phase. The interview
participants were selected through convenience sampling, which involves the
selection of easily accessible and willing participants (Plano Clark & Creswell,
2008). One of the last questions on the survey asked if respondents would be willing
to participate in a follow-up interview, and if so, to provide contact information. If a
person added contact information, then he or she was contacted for a follow up
interview. In addition to those who had completed the survey, architects, building
owners and facility managers were also contacted for building-specific interview
questions and requests.
Interviews. The most prominent method of data collection was through semi-
structured and open-ended interviews with occupants, owners and building managers
via telephone and email. Semi-structured interviews fall somewhere between fully
structured interviews, which are inflexible in nature, and unstructured interviews,
which are not typically preconceived before an interview. As such, questions were
developed before the interviews as a guide, but they remained flexible based on the
occupants’ knowledge of the building and the types of strategies used in the building.
This approach allowed for more flexibility during the interview, and it also allowed
for some responses to emerge that may not have with a more defined set of questions.
The majority of telephone interviews were taped and later transcribed. Extensive
notes were taken for the un-taped interviews. Interview questions varied widely and
were based on individual answers from the survey, or the questions were specific to
the building the occupant represented.
Documents, photographs and architectural drawings. Other means of data
collection included the acquisition of photographs, maps, architectural documents and
plans, and occupant training materials if available.
Establishing validity, reliability and credibility. Rigorous data analysis
techniques were used for all phases of this research, but it is important to point out
that the criteria for measuring validity and reliability differs in quantitative and
qualitative research. In quantitative research, validity means that participants’
assessments are meaningful indicators of the particular construct being measured
© ASCE
(Creswell & Plano Clark, 2011); this definition differs in qualitative research. Four
respected methods were used to validate the qualitative findings to establish
credibility and rigor during the qualitative phase of the study: (1) triangulation –
comparing different types of information (transcripts, documents, survey responses)
(2) rich, thick description of the results; (3) identification of disconfirming evidence
to confirm the accuracy of the findings (Creswell & Plano Clark, 2011; Creswell,
2009); and (4) member checking – a method of providing summaries to participants
to see if the researcher’s interpretation of their conversation was “right” (Marshall, &
Rossman, 2011).
Data Analysis. The recorded interviews were sent to a transcriptionist. After the
transcribed interviews were received, the researcher reviewed the transcripts for
accuracy. After the transcripts were cleaned, they were divided into separate
interview questions before they were imported into the software program for coding.
Some of the questions asked during the interviews varied based on the occupants’
responses to the survey or what type of building strategies had been implemented in
their particular building. Both the open-ended responses from the survey and the
transcribed interviews were analyzed using open and thematic coding in NVivo 10.
Photographs, architectural drawings, and other collected documents were also coded
and grouped by climate type and building code. These additional materials were used
anecdotally during data analysis to help explain findings from either the survey or
interview responses.
RESULTS
As previously discussed, this paper presents findings from two selected

research questions, which revolved around the behaviors of the study participants.
(1) RQ12: Why do occupants interact with the blinds, electric lighting or
thermal controls? Unfortunately, the survey did not specifically ask why occupants
interacted with the electric lighting or thermal controls (study limitation). However,
occupants were specifically asked why they chose to operate the blinds. The results
showed that occupants most frequently adjusted the blinds to decrease the amount of
daylight in their space and to decrease the amount of glare on their computer. The
third most common reason was to increase the amount of daylight.
For the most part, answers to interview questions surrounding why occupants
interacted with other building features were not surprising (too cold, too hot, too
much glare, to let more light in, etc.). One participant worked in a building with red
and green lights, which signaled to occupants when it was ok to open their windows
for natural ventilation (i.e. it was an acceptable temperature outside to let fresh air in
and it wouldn’t affect the heating or cooling system). The occupant said that even
though they knew what the lights meant, they would always open their window
anyway because it was “just too hot…I mean, people will care about their comfort in
this building. And it’s a joke amongst us. You’ve got to keep in mind we’re a bunch of
liberal college-educated people who like the idea of a green building. We had parties
when it was built. But we’re still human animals and so we get really crabby... And
too hot.” Thermal comfort frequently emerged as a motivator for certain behaviors.
© ASCE
(2) RQ13: For what reasons do occupants choose not to interact with high
performance building features? This question could not be answered with any of the
survey results in the quantitative phase. However, there were many comments made
during the follow-up interviews that helped to explain why people chose not to
manipulate high performance building features. After the interviews were coded, four
themes emerged which clarified why people did not interact with the building or
controls. Example quotes are provided for each theme.
• Social concerns (occupants did not want to affect others) and/or the culture in the
office was not conducive to changing thermal or visual conditions
o “normally in my own house I would certainly go ahead and do it [change
the blinds]. But here, I’ll change my own position so it doesn’t affect 20
other people usually - you know, shift positions even though… If the sun’s
only in my eyes, you know, no big deal.”
• “Not my dime” in reference to why they do not care about saving energy
o “… it feels like if you had some individual control [of the thermostat] you
actually would end up with energy saved, because I’ve had my window
open wasting heat a number of times... I had a father who taught me not to
waste energy in the seventies, but the number of times that I have wasted
energy here… it’s because its not my dime, right? It’s because I’m
uncomfortable.”
• Occupants did not understand how to effectively control the features
o “I have a problem remembering which way to tilt the blinds so they
maximize daylight and reduce glare and heat gain so having a reference
for blind positions at the controls or access to a building operation
manual would be useful.”
• Lack of actual control or perceived control (or asked not to touch it by someone)
o “And if you want to close or open the blinds or window …well the
windows are annoying because it's automated…so you can’t control it.”
o “It is just stupid. We all have to stay completely still for 15 whole minutes
before the lights will turn off so that we can actually see outside. Once
they go off, no one moves. It is very annoying.”
DISCUSSION
The follow-up interviews revealed some interesting responses for why

occupants chose not to (or were unable to) interact with building controls and
systems. For instance, one participant said they [all night-time employees] must stay
completely still so the electric lights would turn off since a physical switch was not
provided on the wall. “It is just stupid. We all have to stay completely still for 15
whole minutes before the lights will turn off so that we can actually see outside. Once
they go off, no one can move. It is very annoying.”
This individual worked at a weather station, where they needed to literally
observe the weather. The electric lights inhibited their ability to have ‘night-vision’
for night weather viewing. Because the building team was aiming for LEED
certification, and because the building was completely automated to maintain energy
efficiency, no electric lighting switches were provided to occupants. This was
© ASCE
obviously incredibly inconvenient for occupants who needed night vision capabilities
to perform their job at the weather station. This is a reminder that designers must
understand the use and the intent of the building when implementing high
performance strategies into a building. Not only did occupants truly need control of
the lights to efficiently perform their job, but they were also highly unsatisfied and
frustrated with their environment. This example demonstrates the importance of an
integrated design approach (Brown & Cole, 2006), which takes all necessary use
factors into account during the programming and design phases.
Other reasons occupants chose not to interact with high performance building
features, which emerged from the open-ended survey questions and interview
responses, included the following:
• Social concerns (occupants did not want to affect others) and / or culture
in the office not conducive to changing thermal or visual conditions
• “Not my dime” in reference to why they don’t care about saving energy
• Occupants did not understand how to effectively control the features
• Lack of control or perceived control
Each of these emergent themes relates back to the literature: social influence
(Jain et al., 2013), lack of understanding (Day et al., 2012; Hadi & Halfhide, 2011),
and lack of control (Luo et al., 2014). Additionally, many of the reported motivators
for behaviors related to thermal comfort (or discomfort).
High-performance buildings integrate strategies, such as daylighting and natural
ventilation, which may lead to variable visual or thermal conditions. As such, it is
important that occupants have control of these strategies so visual and thermal
comfort can be maintained. It should also be noted that if occupants are expected to
interact with these building systems to maintain their visual and/or thermal comfort,
then it is of the utmost importance they understand how to control and alter their
conditions. To understand building controls, and to effectively alter conditions to
meet their needs, an occupant may require additional education within the context of
their particular building.
Furthermore, in high-performance buildings, where natural ventilation is often
used as a primary design strategy, the temperature may shift even more than those in
conventional, mechanically controlled buildings. Conventional buildings are
mechanically regulated and aim to provide thermal comfort to only 80% of
occupants, meaning that 20% will most likely be uncomfortable at some point during
the day (ASHRAE Standard 55, 2008). It is argued in the literature that occupants
may have to redefine this “acceptable” range for thermal comfort in high-
performance buildings (Cole & Brown, 2009; Kwok & Rajkovich, 2010) because (a)
it is more difficult to maintain constant temperatures in high-performance buildings,
and (b) space heating, ventilation, and air-conditioning utilize such a large amount of
energy in buildings that expanding our thermal comfort standards would reduce
greenhouse gas emissions and help to conserve energy (Kwok & Rajkovich, 2010).
Some studies have found that occupants may be more willing to tolerate wider
temperature ranges in naturally ventilated buildings when they are given the option of
control (i.e. opening windows themselves) (Cole & Brown, 2009; Humphreys, 2005;
Nicol & Humphreys, 2010).
© ASCE
CONCLUSION
Ultimately, occupant behaviors can either positively or negatively affect a

building’s energy outcomes. Motivators or deterrents to behaviors, such as visual
comfort, thermal comfort, and access (or lack of access) to controls are important
factors for understanding how and why occupants may behave in a certain way. There
are additional factors that may also contribute to occupant behaviors such as social
influences and lack of knowledge of building systems. For instance, occupants are
more likely to alter their conditions, change the blinds, and interact with other
building strategies if they see other occupants exhibit these same behaviors (Jain et
al., 2013) since people’s behaviors often echo what they perceive as the norm
(Goodwin, 2013). As such, it is important to encourage desired behaviors.
There are many ways to nudge occupants into changing their behaviors,
including providing feedback and/or incentives, goal setting, and competitions
(Goodwin, 2013). These strategies have been recommended for behavior change in
both residential and commercial building settings (Brown et al., 2009; Chen et al.,
2012; Jain et al., 2013).
Even if occupants are encouraged to behave in a way that promotes energy
efficiency, a lack of knowledge surrounding building systems may still present a
barrier to these energy efficient behaviors (Sweeney et al., 2013). If energy reduction
is a primary goal for a building, then the building’s occupants must understand how
to interact with the building systems in a way that supports both personal comfort and
energy efficiency.
REFERENCES
Brager, G., & de Dear, R. (2000). A standard for natural ventilation. ASHRAE
Journal, 21-28.
Brown, G.Z, and Cole, J. (2006) Rethinking the design process, retrieved October 4,
2010 from http://www.betterbricks.com/graphics/assets/documents/IED
Brown, Z. B., Dowlatabadi, H., & Cole, R. J. (2009). Feedback and adaptive
behaviour in green buildings. Intelligent Buildings International, 1(4), 296–315.
Chen, H.M., Lin, C.W., Hsieh, S.H., Chao, H.F., Chen, C.S., Shiu, R.S., … Deng,
Y.C. (2012). Persuasive feedback model for inducing energy conservation
behaviors of building users based on interaction with a virtual object. Energy and
Buildings, 45, 106–115. doi:10.1016/j.enbuild.2011.10.029
Cole, R. J., & Brown, Z. (2009). Reconciling human and automated intelligence in
the provision of occupant comfort. Intelligent Buildings International, 1(1), 39–
55. doi:10.3763/inbi.2009.0007
Creswell, J. (2009). Research design: Qualitative, quantitative, and mixed methods
approaches. Thousand Oaks, CA: SAGE Publications, Inc.
Creswell, J., & Plano Clark, V. (2011). Designing and conducting mixed methods
research. (2nd ed.). Thousand Oaks, CA: SAGE Publications, Inc.
Day, J., Theodorson, J., & Van Den Wymelenberg, K. G. (2012). Understanding
controls, behaviors and satisfaction in the daylit perimeter office: A daylight
design case study. Journal of Interior Design, 31(1), 17–34.
© ASCE
Day, J. & Gunderson, D.E. (2014). Understanding high performance buildings: The
link between occupant knowledge of passive design systems, corresponding
behaviors, occupant comfort and environmental satisfaction. Building and
Environment, 84, 114–124. doi:10.1016/j.buildenv.2014.11.003
Galasiu, A. D., & Veitch, J. A. (2006). Occupant preferences and satisfaction with the
luminous environment and control systems in daylit offices: A literature review.
Energy and Buildings, 38(7), 728–742. doi:10.1016/j.enbuild.2006.03.001
Goodwin, S. (2013, July 23). Behavior-based energy efficiency approaches around
the northwest (a table). [Excel spreadsheet]. Retrieved from
https://conduitnw.org/pages/file.aspx?rid=1653.
Guerin, D. A., Kim, H. Y., Brigham, J. K., Choi, S. M., & Scott, A. (2011). Thermal
comfort, indoor air quality and acoustics: A conceptual framework for predicting
occupant satisfaction in sustainable office buildings. International Journal of
Sustainable Design, 1(4), 348–360.
Hadi, M., & Halfhide, C. (2011). Green buildings: Understanding the role of end user
behaviour. Going Green: The Psychology of Sustainability in the Workplace, 31.
Hanqing, W., Chunhua, H., Zhiqiang, L., Guangfa, T., Yingyun, L., & Zhiyong, W.
(2006). Dynamic evaluation of thermal comfort environment of air-conditioned
buildings. Building and Environment, 41(11), 1522–1529.
doi:10.1016/j.buildenv.2005.06.002
Humphreys, M. A. (2005). Quantifying occupant comfort: are combined indices of
the indoor environment practicable? Building Research & Information, 33(4),
317–325.
Jain, R. K., Gulbinas, R., Taylor, J. E., & Culligan, P. J. (2013). Can social influence
drive energy savings? Detecting the impact of social influence on the energy
consumption behavior of networked users exposed to normative eco-feedback.
Energy and Buildings, 66, 119–127. doi:10.1016/j.enbuild.2013.06.029
Kwok, A. G., & Rajkovich, N. B. (2010). Addressing climate change in comfort
standards. Building and Environment, 45(1), 18–22.
doi:10.1016/j.buildenv.2009.02.005
Luo, M., Cao, B., Zhou, X., Li, M., Zhang, J., Ouyang, Q., & Zhu, Y. (2014). Can
personal control influence human thermal comfort? A field study in residential
buildings in China in winter. Energy and Buildings, 72, 411–418.
doi:10.1016/j.enbuild.2013.12.057
Marshall, C., & Rossman, G. (2011). Designing qualitative research. (5th ed.).
Thousand Oaks, CA: SAGE Publications, Inc.
Nicol, F., & Humphreys, M. (2010). Derivation of the adaptive equations for thermal
comfort in free-running buildings in European standard EN15251. Building and
Environment, 45(1), 11–17.
Otter Tail Power Company. (2009, March). Electric space heaters. Retrieved Feb. 9,
2014, from http://www.conservingelectricity.com/ElectricSpaceHeaters.asp
Plano Clark, V., & Creswell, J. (2008). The mixed methods reader. Thousand Oaks,
CA: SAGE Publications, Inc.
Sweeney, J. C., Kresling, J., Webb, D., Soutar, G. N., & Mazzarol, T. (2013). Energy
saving behaviours: Development of a practice-based model. Energy Policy, 61,
371–381. doi:10.1016/j.enpol.2013.06.121
© ASCE
Van Den Wymelenberg, K. (2012). Patterns of occupant interaction with window

blinds: A literature review. Energy and Buildings, 51, 165–176.
doi:10.1016/j.enbuild.2012.05.008
© ASCE
Participatory Energy Management in Building Networks
Mina Rahimian; Daniel Cardoso-Llach; and Lisa Domenica Iulo
Department of Architecture, Penn State Stuckeman School, PA.

E-mail: mxr446@psu.edu; dzc10@psu.edu; ldi1@psu.edu
Abstract
Research on delivering high quality energy-related information on users’

activities and consumption rates signify the effectiveness of such information for
inspiring and motivating users to change their behavior towards more energy saving
ones. However, the issue of making these behavior changes durable and integrated to
one’s lifestyle is still remaining a topic for further investigation. This paper attempts
to encourage new ways of thinking about users’ engagement in the energy
management system of their community-based microgrid by combining
computational means of feedback delivery with an incentive program, which requires
users’ self-organized collaboration and participation in the shared-energy community
endeavor.
Keywords: Participatory; Collective intelligence; Microgrids; Energy

management.
INTRODUCTION
Energy production is typically a regional enterprise, with the majority of

energy produced far from the main areas of demand. This causes tremendous
problems in terms of lack of resiliency and flexibility in handling the ever changing
demands at the users’ end and the continuous changes of the dynamic environment
(Kang, Park, Oh, & Park, 2014 - Farhangi, 2010 - Amin & Wollenberg, 2005 -
Villareal, Erickson, & Zafar, 2014). Microgrids, on the other hand - as localized
energy infrastructures - support resiliency in the electrical grid by exercising greater
control over the production by generating energy close to its point of consumption.
Microgrid integrate various techniques of automation, optimization, pervasive
control and computation on both the supply and demand side (Rahimian, 2015 -
Sherman, 2007 - Paglia, 2011).
At the demand side, microgrids empower the users to interact with the energy
management system to adjust their energy use and reduce their energy costs
(Farhangi, 2010). Energy metering devices and feedback technologies act as a
medium for communication between the grid and the user by making the grid visible
to them and facilitating energy management. The mere display of users’ energy
consumption feedback information, gives users the ability to control their
consumption pattern based on the energy pricing rates throughout the day (Ipakchi &
© ASCE
Albuyeh, 2009 - Kang, Park, Oh, & Park, 2014 - Rahimian, Iulo, & Cardoso Llach,
In Press).
Energy feedback technologies are based on the hypothesis that most people
lack awareness and understanding about how their everyday behavior affects the
environment and therefore limits the consumer’s capacity on deciding to take
conservation actions (Lutzenhiser, 1993 - Froehlich, et al., 2010). In microgrids, it is
expected that displaying energy feedback information at the demand side will
increase users’ knowledge on consumption and consequently leads to managing
energy usage by adopting long lasting conservation behaviors. Several studies
suggest that delivering high quality energy-related information on users’ activities
and consumption rates has the potential to motivate users change behavior towards
more energy saving ones (Yu, Fung, Haghighat, Yoshino, & Morofsky, 2011).
However, feedback delivery alone may not suffice to change people’s consumption
behavior in long term (Hargreaves, Nye, & Burgess, 2010). Thus the issue of how to
make energy-related behavior change durable and integrated to one’s lifestyle
remains a topic for further investigation.
BACKGROUND
Studies of human behavior in the context of energy feedback technologies

have shown that although displaying energy consumption information is necessary
and valuable for increasing awareness and helping consumers control their
consumption, broader psychological, social and cultural patterns of household energy
use must be accounted for in order to encourage long-lasting changes in behavior
(Aune, 2007 - Hargreaves, et al., 2010). This is because metered provisions acquire
meaning after going through each household’s interpretive and discursive lens, point
of view and cultural practices. After energy feedback information is individually
processed it holds persuasive ability and the potential to solve the gap of “energy
illiteracy”, but it does not necessarily inspire users to adopt long-lasting behavior
change.
Over the past fifty years, environmental discourses in the field of human-
computer interaction (HCI), and studies on motivations for environmentally positive
behavior in the field of environmental psychology have been popular subjects for
research. (Goodman, 2009). An extensive literature review on 139 resources by
Froehlich et al on the study and design of environmental HCI1, as the intersection of
environmental psychology and HCI, have explored two distinct approaches
addressing the design and evaluation of feedback technologies which in consequence
have resulted in a profound gap between these two disciplines. These researchers
find this divergence to be the main reason that feedback technologies’ lack success in
promoting long-lasting behavior change (Froehlich, Findlater, & Landay, 2010). This
is while environmental psychologists have largely focused on the effects of energy
feedback information per se and HCI researchers have concentrated on the iterations
1As Froehlich et al assert, environmental HCI is the study and design of eco-feedback technologies
which provides feedback on individual or group behaviors with the goal of reducing environmental
impact (Froehlich, Findlater, & Landay, 2010)
© ASCE
of feedback design and the production of the artifact itself with an emphasis on
understandability, usability and aesthetics (Froehlich, Findlater, & Landay, 2010).
A PARTICIPATORY MODEL FOR ENERGY MANAGEMENT
A growing body of literature suggests that combining feedback delivery on

consumption patterns with other strategies such as goal setting, incentive programs,
economic penalties, etc., is a more effective way of nudging users towards more
responsible energy consumption habits. (Costanzo, Archer, Aronson, & Pettigrew,
1986 - Fischer, 2008 - Froehlich, Findlater, & Landay, 2010 - Hargreaves, Nye, &
Burgess, 2010). These studies have emphasized the complexities of human behavior
and highlighted a body of environmental psychology literature offering techniques
and inspiration on behavior change strategies to guide and/or complement persuasive
energy feedback technologies (Lutzenhiser, 1993 - Hargreaves, Nye, & Burgess,
2010).
Building upon the reviewed literature, this paper recommends a shift in focus
more on the community’s energy use rather than individual energy consumers as the
key unit of analysis (Rahimian, 2015). Toward this end, a conceptual prototype of an
energy exchange system —a collaborative energy sharing network for small-scale
community microgrids— with a diversity of intense energy users, structured on a
collaborative incentive program with interactive and comprehensive energy feedback
information is proposed as a possible solution. The focus is not on directly educating
users about their energy consumption, but rather on fostering cooperative and
energy-saving dynamics by coupling energy feedback technologies and an incentive
program which requires users’ self-organized collaboration and participation in
sharing energy within their community’s microgrid.
The interface system described below is proposed as a way to increase the
possibility of a community microgrid to be energy responsive through its users. In
this system the visualization of energy use through feedback devices, an aesthetically
appealing method for inducing behavior change, is combined with game-like built in
incentives to motivate long term behavior modification. The system seeks to foster
collaboration and participation among users, advancing a new view on energy
consumption as a community endeavor. The “game” can provide targeted incentives
for users of a microgrid to alter their consumption patterns and shape the use of
shared-energy resources, resulting in new patterns of energy responsive collaboration
and participation in the microgrid, linking resiliency to a community’s collective
intelligence.
SYSTEM OVERVIEW
The section below discusses how the proposed conceptual prototype energy
exchange system operates and introduces an interface structured upon an incentive
program that motivates users’ participation.
Conceptual Prototype. The proposed system is a small-scale community microgrid

constituted of a moderate number of interconnected houses with a diverse set of
© ASCE
eneergy users, introduced

i as a method d to address the assum mptions andd principles laid
outt in the prev
vious sectionn. The varieegation of users
u in thiss system plaay an imporrtant
rolee in drivingg the systemm since it intensifies the possibiility of exchanges to take t
placce.
Eneergy Supp ply. For thiis proposal communityy energy iss supplied by a comm mon
souurce of renewwable energgy such as solar, wind power or fuel f A a small-scale
cells. As
miccrogrid the intent is to reach a levvel of indeppendency onn fossil fuells, in whichh the
commmunity’s need
n for ennergy is mostly
m proviided by rennewables. This T is becaause
deppendence onn clean eneergies is a challenging
c task since the amoun nt of renewaable
eneergy harvestted normallly doesn’t match
m the ammount of ennergy consuumed in homes
(Zhhu, et al., 2013).
2 Whiile sharing energy willl keep the supply annd demand rate
balaanced in a microgrid
m c
community, it also servves as an effficient strattegy addresssing
thiss challenge.
Opperation. Inn this connceptual pro ototype thee

opeeration of the
t energy exchange mechanism m
resuults from a pro-ennvironmentaal strategyy
commbining energy feedback technoologies withh
an incentivee structuree promotiing user’ss
parrticipation and collaaboration for savingg
eneergy in the communityy. As a com mputationall
straategy, the energy
e exchhange systeem has twoo
layeers: One is the layerr of compuutation andd
algoorithms whhich technically drives the system,
hanndles com mmunicationn among differentt
houuseholds’ ennergy profiiles and is responsiblee
for the energy transactions. The other o layerr, Fig
gure 1. The translation of the
whiich is mainnly discussed in this paper, is a allgorithms into a user-friendly
y
mplified trannslation of the compuutation layerr
sim interfa
ace
intoo an interacttive user-friiendly interface.
TH
HE INCENT
TIVE-SRU
UCTURED INTERFA
ACE
Thee interface used

u in this strategy is a communnicative webb-service deevice serving as
a medium
m betw
ween the usser, the com mmunity andd the grid, accessed
a froom anywherre in
the house andd outside annd presenteed in manyy
mannifestationss: throughh tablets,, phones,
webbsite profilees and homee dashboardds.
Thee interfacee graphically displlays three
diffferent, butt related set s of datta on the
houusehold’s personal
p ennergy informmation forr
insttance the deebit and creedit-energy accounts
a inn
the “YOURS”” tab (thesee energy acccounts will
be discussed later), the communityy’s general
eneergy informmation suchh as the community-
c
eneergy accouunt in the “OUR” tab, andd Figure 2.. The “YOURS” tab of the interface
© ASCE
recoommended energy conservation suggestions and tips customized d based on the

houusehold’s overall
o connsumption pattern in the “TIPS S” tab. Thhe informaation
dispplayed in these
t tabs help
h users perceive thheir personnal and gro oup benefitss of
makking more efficient ennergy consuumption deccisions and understandd the payofffs of
takiing conservvation actionns.
“YO OURS”: as shown in the image this t tab reppresents hou usehold’s personal
p eneergy
infoormation annd includes several graaphics: 1- thhe energy accounts
a (en
nergy-debit and
eneergy-credit) and the usser’s currennt energy sttatus in term ms of consuumption 2- the
currrent overalll energy effficiency staatus of the house 3- anda the planns display how
h
effiiciently enerrgy is beingg consumedd in differennt locations of
o the housee.
“OURS”: Withh the featurres outlined below, thiss tab attemppts to abstracctly incentivvize
commmunal colllaboration and particippation amoong the useers for savin ng and sharring
eneergy in the communitty. Additionnally it givves general informatioon on how the
commmunity is doing energgy-wise.
• Com mparison: Self and else compparison has
shoowed to bee an effecttive reasonn for takingg
eneergy saving g actions pparticularly when it is
commbined witth feedbacck about performance
p e
(Frooehlich, Findlater,
F & Landaay, 2010)).
Acccordingly, the interface is featu ured with a
visuual evaluattion of eacch householld’s currennt
andd past en nergy behaavior in addition
a too
grap
aphically displaying thhe user’s sttance in thee
oveerall commuunity’s enerrgy consumpption.
• Rew wards and Penalties: Rewards
R annd penaltiess
are consequennce motivatiion techniq ques comingg
afteer a behaviior. Researrch into thee effects off
rew
wards have found thaat people respond too
rew
wards even if i they are nominal
n in nature
n (e.g.,
an acknowledggement of positive
p behhavior). Onee
of the
t main drrivers that makes
m offerring rewardss
a strong motivve for userrs is that thhey becomee
thriived to sett goals forr themselvees to reachh
those rewardss. When thhe goal of winning a Figure 3. The “OURS” tab of the interfa
ace
rew
ward is set, a sense of coommitmentt, a promisee
to behave
b in a specific way
w to attaiin that goall
will thrive in the user and conseqquently thee
commmitment increases the posssibility off
purrsuing sppecific beehaviors (Froehlich,
Finndlater, & Landay, 20100).
“TIIPS”: Forr outliningg any mootivational
straategy it’s im
mportant to highlight and
a lay out
the personal benefits derived
d froom taking
actiion (Ouyangg & Hokao,, 2009). Thee designed
inteerface cleaarly states the perssonal and Figure 4.. The “TIPS” tab
b of the interfac
ce
finaancial benefits of specific conservation
© ASCE
actions in specific situations by means of

graphical visualization under this tab.
HOW IT WORKS
Below is a brief description on how the system works:

The power of the community is supplied by a common source of renewable
energy leading to the resource being shared among community users. Energy-tokens
(eT), each at an established value based on price of electricity and average energy use
per capita, are the suggested currency use in the exchange system. Tokens can be
exchanged in three ways according to user conditions:
o Debit-energy tokens (Debit-tokens): represents the first and main
share of energy that each household receives each month. The amount
of debit-tokens assigned to each household is based upon the number
of family members.
o Credit-energy tokens (Credit-tokens): represents a trading mechanism
for obtaining additional energy-tokens, in which case the household
gets charged by energy-tokens.
o Community-energy tokens (Community-tokens): a system for sharing
energy-tokens were individual households may either sell extra
energy-tokens at the end of the month or purchase energy-tokens if all
debit and credit-tokens are used.
Each month every household receives two constant share of debit-energy
tokens and credit-energy tokens which can be accessed and viewed through the
“YOURS” tab of the interface. The household’s monthly energy usage is tracked
using debit-energy tokens first. In several typical scenarios household energy can be
managed through the use and exchange of debit and credit tokens. If a household’s
consumption goes beyond the limits of the second share of energy (credit-energy
token), in a month, leaving the user in need of extra energy-tokens, the shared
community-token account can be borrowed against, providing household’s access to
additional energy tokens. By using energy-tokens from this shared account, the user
owes the community both energy-tokens and actual money due to the end of the
following month. In order to prevent users from continuously depending on the credit
and community energy accounts for purchasing extra energy, the energy price
follows an ascending pattern.
In a community home to a diverse number of households with different
energy consumption patterns, at the end of the month there are always users which
have been efficient2 and users which have been more or less inefficient3 in terms of
energy consumption. This means every month there are always some users selling
their extra energy-tokens to the community, some buying the energy-tokens, some
2 Their consumption have not exceeded their share of debit-tokens

3 Their consumption reached to the point of using from their credit-tokens or from the community-
tokens
© ASCE
staying in the limits of their debit-energy share and receiving monetary rewards and
some crossing the lines of efficiency and paying back money to the system. By this
conceptual prototype and the introduced framework it is expected that energy and
money transactions constantly occur in the scale of the community and become the
main driver of the energy exchange system. In this scenario the system will be host
to users’ participation in this community endeavor.
While explained the conditions which debit and credit-energy tokens are
being used, a situation might occur that a household’s consumption goes beyond the
limits of the second share of energy (credit-energy token) in a month, leaving the
user in need of extra energy-tokens. In this case a shared community-token account
can be borrowed against, providing household’s access to additional energy tokens.
By using energy-tokens from this shared account, the user owes the community both
energy-tokens and actual money due to the end of the following month. In order to
prevent users from continuously depending on the credit and community energy
accounts for purchasing extra energy, the energy price follows an ascending pattern.
CONCLUSION AND NEXT STEPS
The proposed conceptual model of an energy exchange system operates based

upon a set of principles and fundamentals, illustrating energy consumption and
energy efficiency as a more social and collective process rather than individual.
Researchers argue that it’s not effective nor laudable to aim at reducing users’ life
quality, health and safety while promoting conservation behavior. This while users
automatically adjust their behavior to a good balance between conserving energy and
their own acceptable quality of life (Aune, 2007 - Ouyang & Hokao, 2009).The
proposed sharing system is not aiming at challenge the mentality of “the home as
haven”, as Margarethe Aune expresses, but rather attempts to address it. Rather than
focusing on behavior change specifically the consumption patterns of different users
and the dynamism of their behavioral attributes becomes the basis of the energy
exchange mechanism. Therefore, the energy exchange mechanism operates based on
the user’s tacit knowledge. That is, users perceive the payoffs proposed by sharing,
and they themselves chose and decide to borrow or lend energy to their similarly
situated others in the community based on the energy information provided by
feedback technologies. Ultimately, the HCI and feedback technologies in this system
are integrated in this system for information and communication facilitation.
Therefore the focus is on the users as the smartest component of the system rather
than any so-called smart technological device.
ACKNOWLEDGEMENTS
This research was supported by the Alma Heinz and August Louis Pohland
Scholarship in the Stuckeman School of architecture and landscape architecture.
Professor Ute Poerschke’s support during this research is gratefully acknowledged.
Some information covered in this paper, including a more thorough literature review,
will be published in the forthcoming proceedings for the “Architecture and
© ASCE
Resilience on a Human Scale” conference at the University of Sheffield School of

Architecture, 10-12 September 2015.
REFERENCES
Amin, M., & Wollenberg, B. F. (2005). Toward a Smart Grid. IEEE Power and
Energy, 34 - 41.
Aune, M. (2007). Energy Comes Home. Energy Policy, 5457 - 5465.
Costanzo, M., Archer, D., Aronson, E., & Pettigrew, T. (1986). Energy Conservation
Behavior, The Difficult Path From Information to Action. American
Psychologist , 521 - 528.
Farhangi, H. (2010). The Path of the Smart Grid. IEEE Power & Energy, 18 - 28.
Fischer, C. (2008). Feedback on household electricity consumption: a tool for saving
energy? Energy Efficiency, 79 - 104.
Froehlich, J., Findlater, L., & Landay, J. (2010). The Design of Eco-Feedback
Technology. In Proceedings of the SIGCHI Conference on Human Factors in
Computing Systems, (pp. 1999-2008). Atlanta.
Goodman, E. (2009). Three Environmental Discourses in Human-Computer
Interaction. In CHI'09 Extended Abstracts on Human Factors in Computing
Systems, (pp. 2535-2544).
Hargreaves, T., Nye, M., & Burgess, J. (2010). Making energy visible: A qualitative
field study of how householders interact with feedback from smart energy
monitors. Energy Policy, 6111-6119.
Ipakchi, A., & Albuyeh, F. (2009). Grid of The Future, Are we ready to transion to a
smart grid? IEEE Power & Energy, 52 - 62.
Kang, S. J., Park, J., Oh, K.-Y., & Park, H. (2014). Scheduling-based real time
energy flow control strategy for buildingenergy management system. Energy
and Building, 239 - 248.
Lutzenhiser, L. (1993). Social and Behavioral Aspects of Energy Use. Annual
Review of Energy and the Environment, 247-289.
Ouyang, J., & Hokao, K. (2009). Energy-saving potential by improving occupants’
behavior in urban residential sector in Hangzhou City, China. Energy and
Buildings, 711 - 720.
Paglia, T. K. (2011). Energy Improvement Districts and Local Energy Production
(Master of Regional Planning Dissertation, Cornell University).
Rahimian, M. (2015). A Participatory Approach for Constructing Energy Resiliency.
University Park, PA, USA: Unpublished Master Dissertation. Pennsylvania
State University.
Rahimian, M., Iulo, L. D., & Cardoso Llach, D. (In Press). The Case for a
Collaborative Energy Sharing Network. Submitted for publication at the
proceedings for the "Architecture and Resilience on a Human Scale"
conference at University of Sheffield, School of Architecture, 10-12
September 2015.
© ASCE
Sherman, G. R. (2007). Sharing Local Energy Infrastructure - Organizational Models

for Implementing Microgrids and District Energy Systems in Urban
Commercial Districts (Master of City Planning, Massachusetts Institute of
Technology).
Villareal, C., Erickson, D., & Zafar, M. (2014). Microgrids: A Regulatory
Perspective. CPUC Policy & Planning Division.
Yu, Z., Fung, B. C., Haghighat, F., Yoshino, H., & Morofsky, E. (2011). A
systematic procedure to study the influence of occupant behavior on building
energy consumption. Energy and Buildings, 1409 - 1417.
Zhu, T., Huang, Z., Su, J., Irwin, D., Mishra, A., Menasche, D., & Shenoy, P. (2013).
Sharing Renewable Energy in Smart Microgrids. In Proceedings of the
ACM/IEEE 4th International Conference on Cyber-Physical Systems, (pp.
219-228).
© ASCE
One Size Does Not Fit All: Eco-Feedback Programs Require Tailored Feedback
Ardalan Khosrowpour1 and John E. Taylor1

1
Charles E. Via, Jr. Department of Civil and Environmental Engineering, Virginia
Tech, Blacksburg, VA 24061. E-mail: ardalan@vt.edu; jet@vt.edu
Abstract
As buildings become more efficient and automated, occupants’ role become

more significant to help buildings reach their full energy efficiency potential. There
are several approaches to promote occupant energy efficiency in buildings such as
education, interventions, and eco-feedback programs. Eco-feedback systems have
proven to be more effective than other methods in establishing energy efficient
behavior among peers. Despite the effectiveness of eco-feedback systems, a question
still remains how best to target specific occupants. In this paper, we develop new
metrics to analyze occupant behavior and perform hypothesis tests in an eco-feedback
study conducted in Denver, CO with more than 100 occupants. The results
demonstrate an inconsistent response to eco-feedback programs. Therefore, we
suggest a need for further research to investigate the possibility and potential of
targeted and personalized eco-feedback programs which guide, motivate, and engage
the occupants based on their behaviors, characteristics, and constraints.
INTRODUCTION
Buildings account for more than 40% of energy consumption in the US

(U.S.EIA, 2014). Recent advancements have enabled building energy efficiency (EE)
through automation, equipment upgrade, and material improvement. These
technological enhancements are mostly focused on centrally controllable units such as
HVAC and lighting systems, while miscellaneous energy loads (e.g. plug loads) that
are mostly controlled by occupants, are not as efficient as other major units. The fact
that occupants spend more than 90% of their time indoors (U.S.EPA, 2012) suggests
a great opportunity to help buildings reach their EE potential in full by promoting
changes to occupant behavior. In recent years, various occupant energy efficiency
programs have been developed that range from the traditional EE interventions using
door-hangers (Schultz, Nolan, Cialdini, Goldstein, & Griskevicius, 2007) and mailed
letter (Allcott, 2011) to the real-time, high resolution eco-feedback systems
monitoring individual’s energy consumption using wireless smart meters (Delmas &
Lessem, 2014; Gulbinas, Jain, & Taylor, 2014). The advent of wireless smart meters
in conjunction with cloud-based platform services revolutionized the intervention and
EE programs by providing high resolution, real-time information on occupants’
energy consumption and facilitating the implementation of various psychological
© ASCE
concepts such as social norms (Schultz et al., 2007), goal setting (Erickson et al.,
2013), and gamification (Gamberini et al., 2012) at larger scale at a higher frequency.
Eco-feedback systems that are designed based on social norm effects have
proven to be effective in establishing energy efficient behavior among peers
(Gulbinas, Jain, et al., 2014; Jain, Gulbinas, Taylor, & Culligan, 2013). However,
designing such systems requires a targeted and fair normative comparison
environment to engage all the users regardless of their level of efficiency,
demographics, or equipment. In this paper, we developed two novel energy efficiency
metrics to quantify occupants’ energy-use behavior and conducted multiple
hypothesis tests on occupants’ behavior under various eco-feedback programs. The
results are discussed in detail and future venues of research are suggested to further
increase the efficacy of eco-feedback programs.
BACKGROUND
Numerous energy efficiency eco-feedback studies have been conducted in the

residential and commercial buildings using various psychological approaches to
engage and motivate occupants in taking energy efficient actions. Social norms have
been identified as one of the most psychologically effective approaches in eco-
feedback studies for promoting sustainable energy efficient behavior among
occupants (Gulbinas & Taylor, 2014; Schultz et al., 2007). Allcott (Allcott, 2011)
conducted an analysis on Opower’s energy feedback program which implements a
social norm effect to promote energy efficient behaviors among residential building
neighbors by providing a normative comparison. The outcome suggests that
Opower’s feedback program could reduce the residential sector’s energy consumption
by 2% on average. Delmas and Lessem (Delmas & Lessem, 2014) studied a real-time
eco-feedback system which was designed based on the effect of public and private
energy-use normative comparison. The authors reported that public disclosure of
occupants’ energy consumption results in a significantly higher energy savings
compared to private eco-feedback reports. Schmitt (Schmitt, 2014) conducted a study
on more than 600 residential buildings with the goal of examining the effect of
various normative comparison door-hanger information on occupants’ behavior. The
report showed that Opower’s model where occupants are compared to their most
efficient neighbors to be the more effective methodology to reduce occupants’ energy
consumption.
Despite the extensive research conducted on eco-feedback systems and social

norms in the residential sector, this field is relatively unexplored in the commercial
sector. Bradely et al. (Bradley, Leach, & Fudge, 2014) studied the effect of social
norms in incentivizing energy reduction in organizations. Their monthly energy
feedback coupled with multiple interviews and surveys brought them to the
conclusion that the level of descriptive norm has an effect on the energy efficiency of
participants. Gulbinas et al. (Gulbinas, Jain, et al., 2014) developed a real-time cloud-
based eco-feedback system that engaged an individual, a social, and an organizational
network effect (i.e. normative comparison among employees). They reported that
© ASCE
employees who were part of an organizational network group significantly saved

energy compared to the baseline. In spite of advancements focused on social norm
application in eco-feedback systems and their effect on occupant energy
consumption, social norms are not the complete answer to the occupant energy
efficiency challenge as there is still room for improvement of these programs.
Currently, the aforementioned feedback systems in commercial buildings focus on the
energy intensity values (i.e. Wh) to impose the social norm, and motivate the
occupants to save energy. These programs treat all the occupants uniformly and
provide them all with the same type of feedback, while occupants’ different level of
efficiency, demographics, appliances, and equipment make their behavior and energy
efficiency potentials dissimilar. The effectiveness of targeted and tailored feedback
programs have already been established in other areas of research, such as health
(Noar, Benac, & Harris, 2007; Smeets, Kremers, De Vries, & Brug, 2007). This
indicates that there is potential for a more advanced set of metrics for occupants’
behavior analysis and tailored normative comparison systems that maximizes the
effectiveness and sustainability of energy efficiency eco-feedback.
In this paper, we utilize empirical data collected in an eco-feedback study

conducted in a commercial building located in Denver, CO with more than 100
employees to analyze the effect of behavioral differences among occupants. We
tested multiple hypotheses to evaluate the effectiveness of proposed behavioral
metrics with respect to occupants’ energy consumption patterns. Furthermore, a new
methodology is proposed to maximize the effectiveness of eco-feedback systems
through providing targeted feedback and implementing new normative comparison
measures. In the following section, algorithms and hypothesis tests are described.
METHODOLOGY
Due to a high variability of equipment types and counts at each workstation in

office buildings, using energy intensity as a measure of normative comparison will
favor employees with less equipment connected to their workstations. Occupants who
are ranked higher in these systems are not necessarily the most energy efficient in a
commercial building. In this section, we introduce two new metrics that reduce the
dependence of occupants’ normative comparison on equipment variability. It analyzes
employees’ behavior not only based on their energy consumption, but also based on
energy savings and extra shared resources (i.e. HVAC and lighting in this case)
energy-use. Multiple hypotheses evaluate the effectiveness of the metrics on
occupants’ behavior analysis and identifying the most common behavior behind
occupants’ energy efficiency.
Off-working Hours Energy Saving. We build upon previous work on occupants’

behavior classification, and disaggregate occupants’ behavior based on their energy-
use patterns into working days vs. non-working days and working hours vs. off-
working hours (Gulbinas, Khosrowpour, & Taylor, 2014). We developed an
algorithm using MATLAB to evaluate the occupants’ energy savings when they are
not at work (i.e. off-working hours and non-working days). It searches for the
© ASCE
ind
dividuals’ lowwest level of
o energy-use in their ennergy consuumption histtory or uses
the average offf-working ho ours energy value of all employees (whichever is less) and
meaasures the off-working
o hours energ gy savings bby penalizing the excesssive energy
connsumption with
w respect to t their best performance
p e, and rewarrds based on the amount
of energy
e savedd with respeect to work-h hours averagge energy coonsumption. In order to
commpare the reesults and consider
c thee individualss’ energy effficiency pootential, the
norrmalized data is used to reward occu upants’ enerrgy saving effforts. Fig. 1 depicts an
oveerview of thee off-workinng hour energ gy saving reeward algoritthm. The summation off
eacch individuall’s efficiencyy reward (i.ee. the green area) and innefficiency ppenalty (i.e.
the red area) demonstrate
d their energgy efficienccy status com mpared to ttheir peers.
Furrthermore, the normalizzed data red duces the eeffect of equuipment varriability on
occcupants’ EE ranking and provides a fairf compari son measuree.
Fig 1. An overview
o off off-working hour enerrgy saving ccalculation p
process
Sha ared Resourrces Energy y Saving. Un nfortunatelyy in eco-feeddback studiess conducted

in commercial
c buildings,
b th
here is little emphasis onn the impact of occupantts’ behavior
regarding shareed resourcess EE. The implementat
i tion of wireeless smart m meters at a
worrkstation lev vel facilitates the identiffication of ooccupants’ w working hourrs based on
theiir energy coonsumption. Assuming that t shared resources ennergy consuumption are
neccessary and irreducible
i by
b occupantss during officcial workingg hours (e.g. 8am-5pm),
it raises the isssue that the employees who w stay aft fter hours inn the buildinng while the
building is nott fully occu upied are reesponsible fo for a higherr percentagee of energy
connsumption. An A extra ho our of afterr-hour workking could result in m more shared
resoources energy consump ption than an a employeee’s workstaation utilizattion for an
entiire working g day. Therrefore, deveeloping a nnormative ccomparison system to
demmonstrate thee impact of occupants after a hour w
working on shared resourrces energy
connsumption notn only willl increase the t occupannts’ awareneess, but alsoo will help
emp ployees to become fu urther energ gy efficient. We develloped a meetric which
pennalizes occup pants’ energy efficiency ranking by multiplyingg their extra hhours in the
building by an average shaare of lightin ng and HVA AC system ennergy use peer occupant
calcculated based on ASHRA AE guidelin nes (ASHRA AE, 2004).
© ASCE
Daily Energy-use Intensity. Daily energy consumption is the most common metric
used in eco-feedback systems as a means of normative comparison. This metric is
necessary but not sufficient to establish sustainable energy efficient behavior. Thus,
we calculate the daily average energy-use as a part of our comprehensive metric.
Hypotheses. We re-analyzed the data collected from a previous eco-feedback study

in a commercial building with more than 100 employees to investigate the effect of
eco-feedback systems on occupants’ behavior. However, after data cleansing and
classification steps only 75 occupants’ energy data could be included in this study.
Gulbinas and Taylor (2014) conducted an eco-feedback study by categorizing
occupants into three groups: individuals (Group X), organizational network (Group
Y), and social group (Group Z). The authors refer the reader to (Gulbinas & Taylor,
2014) for further information on the details of previous study. Gulbinas and Taylor
(2014) reported that group Y was the only group that significantly changed their
behavior compared to the baseline. The report reveals that group X saved more
energy on average, however, the results were not statistically significant.
We suspected that the organizational network normative effect might not be
the only reason behind occupants’ energy saving. Thus, we ranked our occupants
based on the metrics introduced in the previous sections and divided them into three
quintiles (i.e. top, medium, and low ranks) with an average of 25 occupants in each.
In total, we tested 9 hypotheses (3 quintile × 3 metrics) to study the top-ranked
occupants’ behavioral change under the effect of the eco-feedback system. Because
our data did not pass the test for normality (i.e. Kolmogorov-Smirnov), we adopted a
log transform approach and employed a non-parametric statistical method—the
Wilcoxon paired test—to evaluate the hypotheses below.
Hypothesis 1, 2, 3: Occupants who are ranked in the top, medium, and bottom
quintile of “Energy Savings” (Group A, B, C), reduced their energy consumption
after the eco-feedback program was launched.
Hypothesis 4, 5, 6: Occupants who are ranked in the top, medium, and bottom
quintile of “Off-Working Hours Energy Savings” (Group D, E, F), reduced their
energy consumption after the eco-feedback program was launched.
Hypothesis 7, 8, 9: Occupants who are ranked in the top, medium, bottom quintile of
“Shared Resources Energy Savings” (Group G, H, I), reduced their energy
consumption after the eco-feedback program was launched.
RESULTS & DISCUSSION
The results of Wilcoxon paired test on energy saving performance of group A-

I are reported in Table 1. At a confidence level of 95%, and a marginal confidence
level of 90%, evidence is weak to reject all null hypothesis, except the ones
associated to hypothesis 1, 4, 7, and 8. Furthermore, the occupants’ distribution
shows that at least 30% of the group X, Y, and Z’s members have overlap with the
members of group A-I. This fact implies that irrespective of the occupants’ eco-
feedback program, their energy-use behavior changes were either significant or
© ASCE
marrginally signnificant, and

d their behavvioral patterrn could be captured annd analyzed
usinng our new wly developeed metrics. An implicaation of thiss inconsistennt response
undder the unifiied eco-feeddback prograams is that a single ecoo-feedback ssystem does
not fit all occup
pants. Thereefore, there is
i room for further proggress in deterrmining the
effeect of targeeted and peersonalized eco-feedbacck systems on occupannts’ energy
effiiciency.
ver performaance relative to baseline and control group

Table 1. Energy sav
Among occupants who saved energy (i.ee. Group A A, B, C, E, and I) the

med dium and bo ottom group ps (i.e. B, C,, E, and I) ddid not show
w significantt changes in
theiir behaviors and instead d group A reeduced their energy connsumption a statistically
signnificant amoount. Occupaants who saaved energy by turning off their apppliances (or
usinng standby mode) durin ng off-work king hours, did not siggnificantly chhange their
eneergy consum mption behav viors (i.e. group
g E), hoowever, thee top off-woorking hour
eneergy savers (i.e.
( group D who savess the most bby turning off or using tthe standby
mode in appliaances after work
w hours) had a margginally signiificant increase in their
eneergy consum mption by 6..4%. One of the impliccations of a marginallyy significant
respponse to the off-working g energy sav ving metric ccould be the existence off a common
undderlying mo otivator amo ong occupan nts which, regardless of their ecco-feedback
pro
ogram, encou uraged them to increase the level off their energyy consumptiion after the
pro
ogram was laaunched. Th herefore, thiss energy effificiency metrric could bee used as an
effeective analysis approach h to the occu upants’ enerrgy-use behavior disagggregation in
an office enviro onment. Thiis provides the t potentiall to performm root-cause analyses to
deteermine the occupants’ inefficient behaviors and habitss. The aforementioned
appproach enablles us to rank k occupants based on thhe normalizeed value of ttheir energy
sav
vings (indepeendent of equ uipment typee or count) rrather than thheir energy uusage.
The occcupants who o worked leess after hoours showedd significantt change off
behhavior relativ
ve to their baaseline in thee top and meedium quanttiles (i.e. grooup G & H),
howwever, the ennergy consum mption increeased rather than decreaased in those cases. This
hass important implicationss for potentiial EE and work load ttrade-offs w which might
© ASCE
explain occupants’ inefficiency. However, more research on this topic needs to be

undertaken before the trade-off between occupants’ energy efficiency, work load, job
duty, and productivity is more clearly understood. Obtained results indicate that
people ranked in the top energy saving group (i.e. A) have the highest potential to
significantly reduce their energy consumption (regardless of their eco-feedback
program) while the results show a marginally significant energy increase for groups
D, G, and H. Statistically significant behavior changes for occupants enrolled in
various eco-feedback programs emphasizes the need for a targeted and personalized
eco-feedback system. Moreover, inconclusive behavior changes observed in the other
categories (i.e. group B, C, E, F & I) or increases in the energy consumption (i.e.
group D, F, G & H) with respect to the baseline calls for further in depth behavioral
analysis of occupants. Despite insightful results obtained in this study, in the future
hypothesis tests can be conducted more precisely by implementing mixed effect
regression models (MRM). Also, this study was conducted based on 75 occupants’
energy-use data. Including more data points from occupants with various
characteristics, behaviors, and demographics may improve and extend these findings.
CONCLUSION
New behavioral analysis and ranking systems are needed in order to better
understand occupants’ reactions to eco-feedback programs and to improve the
normative comparison methodologies implemented in current eco-feedback research
in commercial buildings. In this paper, we develop new metrics to analyze occupant
behavior and performed hypothesis tests based on data collected in an eco-feedback
study conducted in a commercial building in Denver, CO. Inconsistent responses
under the unified eco-feedback program indicate that a single eco-feedback system
does not fit all occupants. These findings suggest that there is potential to improve the
efficacy of EE programs by providing occupants with targeted feedback.
AKNOWLEDGMENTS
This material is based upon work supported by a Department of Energy

(DOE) Building Innovators Grant and the National Science Foundation (NSF) under
Grant No. 1142379. Any opinions, findings, and conclusions or recommendations
expressed in this material are those of the authors and do not necessarily reflect the
views of DOE or NSF. The authors would also like to thank the Alliance for
Sustainable Colorado for hosting the study and Tugce Duzener for data analysis.
REFERENCES
Allcott, H. (2011). Social norms and energy conservation. Journal of Public

Economics, 95(9-10), 1082-1095. doi: 10.1016/j.jpubeco.2011.03.003
ASHRAE. (2004). Standard 90.1-2004, Energy standard for buildings except low rise
residential buildings. American Society of Heating, Refrigerating and Air-
Conditioning Engineers, Inc.
© ASCE
Bradley, P., Leach, M., & Fudge, S. (2014). The Role of Social Norms in
Incentivising Energy Reduction in Organizations.
Delmas, M. A., & Lessem, N. (2014). Saving power to conserve your reputation? The
effectiveness of private versus public information. Journal of Environmental
Economics and Management, 67(3), 353-370. doi:
10.1016/j.jeem.2013.12.009
Erickson, T., Li, M., Kim, Y., Deshpande, A., Sahu, S., Chao, T., . . . Naphade, M.
(2013). The dubuque electricity portal: evaluation of a city-scale residential
electricity consumption feedback system. Paper presented at the Proceedings
of the SIGCHI Conference on Human Factors in Computing Systems.
Gamberini, L., Spagnolli, A., Corradi, N., Jacucci, G., Tusa, G., Mikkola, T., . . .
Hoggan, E. (2012). Tailoring feedback to users’ actions in a persuasive game
for household electricity conservation. Persuasive Technology. Design for
Health and Safety, 100-111.
Gulbinas, R., Jain, R. K., & Taylor, J. E. (2014). "BizWatts": A modular socio-
technical energy management system for empowering commercial building
occupants to conserve energy. Applied Energy.
Gulbinas, R., Khosrowpour, A., & Taylor, J. E. (2014). Segmentation and
Classification of Commercial Building Occupants by Energy-Use Efficiency
and Predictability. IEEE Transaction on Smart Grids.
Gulbinas, R., & Taylor, J. E. (2014). Effects of real-time eco-feedback and
organizational network dynamics on energy efficient behavior in commercial
buildings. Energy and Buildings, 84(0), 493-500. doi:
http://dx.doi.org/10.1016/j.enbuild.2014.08.017
Jain, R. K., Gulbinas, R., Taylor, J. E., & Culligan, P. J. (2013). Can social influence
drive energy savings? Detecting the impact of social influence on the energy
consumption behavior of networked users exposed to normative eco-feedback.
Energy and Buildings, 66, 119-127. doi: 10.1016/j.enbuild.2013.06.029
Noar, S. M., Benac, C. N., & Harris, M. S. (2007). Does tailoring matter? Meta-
analytic review of tailored print health behavior change interventions.
Psychological bulletin, 133(4), 673.
Schmitt, J. (2014). Normative Social Influence and the Moderating Role of Group
Identification: A Field Experiment on Household Electricity Consumption.
Schultz, P. W., Nolan, J. M., Cialdini, R. B., Goldstein, N. J., & Griskevicius, V.
(2007). The constructive, destructive, and reconstructive power of social
norms. Psychological science, 18(5), 429-434.
Smeets, T., Kremers, J., De Vries, H., & Brug, J. (2007). Effects of tailored feedback
on multiple health behaviors. Annals of Behavioral Medicine, 33(2), 117-123.
U.S.EIA. (2014). Annual Energy Outlook 2014 Early Release Overview.
U.S.EPA. (2012). The Inside Story: A Guide to Indoor Air Quality. Retrieved from:
http://www.epa.gov/iaq/pubs/insidest.html#Refguide
© ASCE
Development of Non-Intrusive Occupant Load Monitoring (NIOLM) in

Commercial Buildings: Assessing Occupants’ Energy-Use Behavior at Entry
and Departure Events
Hamed Nabizadeh Rafsanjani, S.M.ASCE1; Changbum R. Ahn, Ph.D., M.ASCE2;

and
Mahmoud Alahmad, Ph.D.3
1
Ph.D. Student, Construction Engineering and Management, The Durham School of
Architectural Engineering and Construction, University of Nebraska–Lincoln, 113
NH, Lincoln, NE 68588-0500. E-mail: hnabizadehrafsanj2@unl.edu
2
Assitant Professor, Construction Engineering and Management, The Durham School
of Architectural Engineering and Construction, University of Nebraska–Lincoln, 113
NH, Lincoln, NE 68588-0500. E-mail: cahn2@unl.edu
3
Associate Professor, Architectural Engineering, The Durham School of Architectural
Engineering and Construction, University of Nebraska–Lincoln, 206A PKI, Omaha,
NE 68182-0681. E-mail: malahmad2@unl.edu
Abstract
Occupancy related energy-use behavior has a significant influence on building

energy consumption. Variations and uncertainties in occupants’ energy behavior
provide the main obstacle for researchers to analyze and predict the impact of
occupant behavior on building energy consumption since commercial buildings often
have such a large number of residents with unique energy-use patterns. However, this
paper hypothesized that individual occupants have their own individual energy
consumption patterns and will typically follow such patterns consistently over time.
Thus, this research studies occupant behavior in an office environment to examine
whether commercial building’s occupant’s energy-use behaviors are consistent over
time. In particular, this research focuses on delay intervals between the occupancy
entry/departure events and the beginning/end of the occupant’s energy-consuming
behaviors. Occupants’ entry and departure events were detected by passively
capturing Wi-Fi packets from occupants’ smartphones while plug-load monitoring
detected the beginning/end and quantity of energy use. Results from a four-week long
period of tracking individual occupants confirm that occupants use a consistent
pattern of starting and ending their energy-use behaviors. Based on these results, this
research supports a framework of non-intrusive occupant load monitoring (NIOLM)
for tracking occupant-specific energy consuming behaviors in commercial buildings.
In the NIOLM framework, the process of tracking each occupant leverages existing
Wi-Fi networks, and building energy-monitoring data aggregates energy-
consumption data for occupants. Thanks to this study’s findings, NIOLM provides a
new opportunity for current industry and research efforts to track occupants’ energy-
consuming behaviors at a minimal cost.
© ASCE
INTRODUCTION
Residential and commercial buildings currently account for 40% of the total
annual energy consumption in the U.S., making them the largest consumer of the
country’s total energy use (U.S. Department of Energy 2011). In particular, the
commercial sector consumes half of all buildings’ energy use (U.S. Department of
Energy 2011), and its energy demand continues to grow faster than other energy-use
sectors (e.g., industry and transportation) (U.S. Energy Information Administration
2014). In particular, more than 80 percent of buildings’ total energy is typically
consumed during the operational phase (United Nations Environment Programme
2007). Therefore, maximizing energy saving during the operational phase when the
building is actually in use is critical.
A growing body of research posits that changing occupants' behaviors is one
of the most cost-effective approaches to achieving energy savings and offers a great
potential for contributing to energy savings (Rieur and Alahmad 2014). Built
environment’s energy use is highly connected to the energy-use behavior of its
occupants, and commercial buildings are estimated to have a potential 6-38% energy
reduction through occupant behavior changes (Azar and Menassa 2013; Kavulya and
Becerik-Gerber 2012; Meier 2006; Staats et al. 2000). Occupant-driven usage in
commercial buildings can also lead to increased waste—excess energy consumption
can account for up to 150% more energy use than necessary (Clevenger et al. 2014).
Unfortunately, such research into occupants’ behavior change significantly relies on
the data availability of occupant-specific energy consumption, but an overview of the
current literature indicates the difficulties in tracking occupant-specific energy
consumption due to the diversity of occupants and complexity of tasks in commercial
buildings (Gulbinas et al. 2015).
To address this issue, our previous study (Chen and Ahn 2014) presented a
concept of non-intrusive occupant load monitoring (NIOLM) that estimates occupant-
specific energy consumption by linking building energy-load variations with
occupancy-sensing data. However, individualities and uncertainties of occupants’
energy use behavior pose a challenge in the pursuit of this approach’s development.
For example, the time lags between occupant’s entry/departure events and the energy-
load variations caused by him/her may vary by occupants and also by each entry and
departure event of an occupant. As part of our effort to address such issues, this study
aims to analyze the repeatability of occupants’ energy-use behaviors, particularly
those related to time lags between occupancy events and energy-use starting/ending
events. This research will provide a basis for developing an algorithm to establish the
linkage between datasets collected by two different systems: occupancy sensing and
building energy management systems (BEMS).
NIOLM MOTIVATION AND FRAMEWORK
Conventional techniques for monitoring occupant-specific energy

consumption typically have estimated the energy consumption of individual
occupants in commercial built environments by using individual plug load meters, but
such methods are not applicable in practice as they demand a large capital
© ASCE
investment. Alternatively, a fundamental pursuit to monitor occupant-specific energy

consumption could be to extend the concept of non-intrusive load monitoring from
individual appliances to individual occupants. In commercial buildings, a single
meter is typically installed at the main electrical service to measure the aggregate
energy consumption of that entire building with all of its occupants. Non-intrusive
load monitoring techniques are then applied to disaggregate the energy data to
identify large appliances Although the current status of such techniques indicates
their effectiveness in residential buildings (Zoha et al. 2012), their effectiveness in
commercial buildings is quite limited due to the number and abundance of similar
appliances in use simultaneously (e.g., personal computers).
In this context, our previous paper (Chen and Ahn 2014) presented the concept
of non-intrusive occupant load monitoring (NIOLM), which extends the concept of
non-intrusive load monitoring to monitoring occupant-specific energy consumption.
The foundational pursuit behind NIOLM is to link energy-consumption data with
occupancy-sensing data. This pursuit is made possible by the current status of
occupancy-sensing technologies, which provides opportunities to economically
monitor individual occupants and their individual energy consumption (Gulbinas and
Taylor 2014). Though there are currently many solutions for occupancy sensing (e.g.,
temperature, motion, CO2 sensors), Wi-Fi based tracking is considered an easy-to-
deploy and more affordable option (Lassabe et al. 2006). Since there are usually
multiple overlapping access points (APs) in commercial buildings, Wi-Fi-based
occupancy sensing can work at a minimal cost. For this reason, our NIOLM framework
leverages existing Wi-Fi networks to track each occupant and links these data with
aggregated energy-consumption data provided by BEMS (see Figure 1). Correlating the
results leads to disaggregated building energy-consumption data and provides
opportunities for monitoring each occupant’s energy consumption.
In brief, the approach would work like this: The Wi-Fi network detects
occupancy entry and departure events based upon the connection and disconnection
of occupants’ smartphones to APs. The event detection is then associated with
occupants’ smartphone MAC addresses stored in the database in order to determine
who entered/departed the building. The energy-load data for the building is also
tracked, and variations in energy load are then correlated with the occupancy events
to monitor the energy-use of individual occupants (see Figure 1). Our previous study
(Chen and Ahn 2014) highlighted this approach’s feasibility by demonstrating that
the events of Wi-Fi connections and disconnections in a commercial building show
promise as a viable indicator for the starting and ending of individual occupants’
energy-consuming behaviors, respectively.
© ASCE
Figure 1. NIOLM
N fram
mework
RE
ESEARCH OBJECTIV
O VE AND HY
YPOTHESE
ES
In orderr to facilitatee the propossed NIOLM framework’’s formationn, this paper

exaamines the occupants’ beehavior at th heir entry/depparture evennts. Figure 2 shows that
therre is typicaally an entraance delay interval bettween the ooccupancy eentry event
detection and the
t beginnin ng of the eneergy-consum ming behavior. The Wi--Fi network
detects the occcupancy entrry event in the buildingg. Then, afteer a few mooments, the
occcupant will create an ad dditional loaad by startinng an energgy-consuming behavior,
and
d the energy y-load increaase obtained d through thee meter of tthe building will detect
the starting evvent in the energy-conssuming behaavior. Thereefore, there is a delay
betwween an occcupant’s enttry event to the buildingg and the beeginning of the energy-
con
nsuming beh havior at his//her work deesk. Similarlly, there is aalso a delay interval for
dep
parture eventts (see Figurre 2).
Understtanding such h time lags between
b occuupancy deteection eventss (i.e., entry
and
d departure) and buildin ng energy loaad variationns will be criitical in devveloping the
presented NIOL LM framew work. We hy ypothesize thhat such tim
me lags of ann individual
occcupant are quite
q consisttent over timme due to thhe repeated patterns of occupants’
eneergy-consum ming behaviors. The validation
v off this hypoothesis willl help link
occcupancy detection even nts with eneergy-load vaariations in order to estimate the
eneergy-load chaange caused d by an indivvidual occupaant (see Figuure 1).
Therefo
ore, this stud dy tests the fo
ollowing hyppotheses:
© ASCE
Hyypothesis 1. There is not a significaant variabilityy in the delaay interval bbetween the
occcupant’s entrry event and d the start off his/her enerrgy-consum
ming behaviors since the
occcupant practtices a consistent delay pattern reppeated over time. H0: T There is no
stattistically siggnificant diifference am mong differrent delay intervals bbetween an
occcupant’s entrry events an nd his/her en nergy consum mption. H1: There is a statistically
signnificant diffference among differentt delay interrvals betweeen an occuppant’s entry
eveents and his/h her energy consumption
c n.
Fiigure 2. Sam
mple of delaay interval
Hyypothesis 2. There is not a significaant variabilityy in the delaay interval bbetween the
endd of energy--consuming behaviors and a the occcupant’s depparture evennt since the
occcupant practtices a consistent delay pattern reppeated over time. H0: T There is no
stattistically sig
gnificant diifference ammong differrent delay intervals bbetween an
occcupant’s energy-consum ming behaviors and his/hher departuree events. H1: There is a
stattistically sig
gnificant diffference am
mong delay intervals bbetween an occupant’s
eneergy-consum ming behavioors and his/heer departure events.
EX
XPERIMEN
NT DESIGN
N AND MET
THODOLO GY
An exp periment waas designed d and conduucted to innvestigate thhe research

hyppotheses. Thhis experimen nt was carried out over a four-weekk period of time on five
parrticipants at an office space
s locateed on the m main campuss of the Unniversity of
Neb braska-Lincooln (UNL). The office space
s is locaated at the ccenter of Nebbraska-Hall
buiildings and has
h three doo ors, each of which openns to a separaate corridor.. Therefore,
the participantss had variouss options forr accessing tthe office. BBefore the exxperiment, a
pre-survey wass done to dettermine the condition
c off the particippants’ smartpphones. The
© ASCE
survey results indicated that participants always leave their smartphone’s Wi-Fi on,
which provided an opportunity for us to capture their Wi-Fi packets. During the
experiment, the energy consumption of each participant was tracked at his/her work
desk by installing a plug-in level meter. The meter—“watts-up?.net”—collected the
real-time electrical energy-load data with 1-second interval resolution. In addition, a
local access point was set up to track the entry/departure events of the participants. A
database was also created to store each participant information from the
entry/departure events and associated energy behavior.
Since we needed to test whether an occupant has a consistent delay interval
pattern repeated over the experiment, we divided collected data for delay interval of
entry/departure events for each participant into two groups: 1st 2-week and 2nd 2-
week. Comparing the two groups would help to find whether individual occupants’
delay interval patterns repeat over time. An appropriate statistical comparison
between the data of the two groups could then address the hypotheses. For this
reason, a two-sample t-test was chosen as the scenario constructed for testing our
hypotheses.
RESULTS AND DISCUSSION
Figure 3 presents the scatter plot for all entry and departure delay intervals of
Participant #1. This figure, as a sample of raw data, shows that the variability across
delay intervals for entry/departure events is low, which indicates that delay intervals
repeat over time. It is noteworthy that across all the raw data, the variability for delay
intervals of all five participants is low; a high variability would likely indicate that a
participant does not have a specific delay interval repeated over time. Furthermore,
figure 4 presents the boxplots for entry and departure delay intervals for two groups
of data for Participant #2. These boxplots, as a sample, show there is no significant
difference between the means of the two groups of data. The boxplots also shows that
the variance of the two groups of data is approximately equal.
Table 1 also presents the results of conducting a t-test on entry/departure
delay intervals for each participant. Based on the results, we fail to reject our null
hypotheses for all five participants, which means that there is no significant
difference among delay intervals for individual participants. The results of the t-test
would therefore indicate that individual participants have their own delay interval
patterns repeated over time.
The specific delay interval pattern for each occupant could be a critical detail
for determining the start/end of energy-consuming behavior. Since the developed
NIOLM framework is founded on occupancy-detection events and there are a large
number of occupants in commercial buildings with different entry and departure
times, it is possible that some entry or departure events happen at the same time. In
such cases, knowing each occupant’s specific delay interval as it is repeated over time
would be helpful in correlating various load variations with various entry/departure
events to find load variations created by each individual occupant.
© ASCE
Fig
gure 3. Scattter plots forr delay interrvals of Parrticipant #1
Figure
F 4. Box
x plots for delay
d intervvals of Participant #2
Tabble 1. Two-S
Sample t-Teest for Delay y Intervals
Enttry Event Dep parture Eveent
Parrticipant
t-Statistic t-Crittical 1 t-Statistiic Critical 1
t-C
#1 1.085290 2.0599538 1.0096700 2..063898
#2 0.015134 2.0555552 0.959831 2..051831
#3 1.188885 2.0599538 0.8450555 2..079613
#4 2.072843 2.0933024 1.1168122 2..093024
#5 0.132360 2.0799613 1.7303244 2..063898
1
α = 0.05
Based on
o the storedd energy datta in the dattabase, we aalso observeed that each
parrticipant typpically incrreases/decreaases the looad by a consistent amount at
entrry/departuree events. Tab
ble 2 shows the amountts of energy load increaase/decrease
and
d the corresponding freequency for each particcipant. Suchh findings inndicate that
ind
dividual occu upants have their own specific eneergy-use related behavior repeated
© ASCE
oveer time at the

t start/end d of their energy-consu
e uming behaavior; such energy-use
beh
haviors are critically imp
portant in dev
veloping NIIOLM.
Tabble 2. Frequ uency of Load Increasee/Decrease

Occupaant
Evvents #1
# #2
2 #3 ##4 #5
1 2
L F L F L F L F L F L F L F L F
15 10 2 10 2 2 10 7 1 2 5 10 3 2 10
Entry
E 9
0 0 5 0 5 8 0 2 0 5 2 5 9 5 0
Deppartur 15 10 2 10 2 3 10 6 1 2 6 10 3 2 10
5
e 0 0 5 0 5 2 0 8 0 5 1 5 4 5 0
1 Lo
oad increase at entry event, orr, load decrease at departure event
2
Frequency (%)
Continuuous occupaancy sensing g with Wi-Fii infrastructuure and corrrelating the

occcupancy sen nsing resultss with enerrgy load vaariation couuld also be helpful to
undderstanding the generall energy-usee behavior oof individuaal occupantss. Figure 5
presents an ex xample of howh NIOLM M can provvide real-tim me energy-uuse data of
ind
dividual occuupants. This example sh hows the occcupant is noot in his/herr work desk
witthin 10:00-111:00AM, butb he/she left l some aappliances oon. Thereforre, we can
undderstand thaat this occu upant occasionally has energy-waasting behavviors. Such
eneergy-use behhavior tracking provides a great oppportunity tto improve behavior is
pairred with pro
oviding effecctive real-tim
me energy feeedback to occcupants.
Figure 5. Real--time trackiing of energgy consumin

ng behaviorr
CO
ONCLUSION
This papper introducced and testeed the conceppt of monitooring the delay intervals
betw
ween occupants’ entry/d departure events and theeir energy coonsumption. The results
ind
dicated that individual
i occcupants hav ve unique deelay intervall patterns reppeated over
tim
me. Such speccific delay innterval patteerns can servve as the fouundation for developing
the NIOLM fraamework, whichw will serve as an eefficient toool for gatherring critical
© ASCE
data about individual occupants’ energy consumption. By providing a real-time

energy-use behavior tracking system, the NIOLM could foreseeably make a great
contribution to research working on improving occupancy energy-use behaviors in
commercial buildings. Furthermore, since there is no need to install new
infrastructure to implement the NILOM, this tool is especially economic for
commercial buildings.
ACKNOWLEDGMENT
This work was financially supported by the Research Council

Interdisciplinary Grant Award (#26-1122-9001-002) of the UNL. Any opinions,
findings, conclusions, or recommendations expressed in this paper are those of the
authors and do not necessarily reflect the views of the UNL Research Council.
REFERENCES
Azar, E., and Menassa, C. C. (2013). “Influence of Social Sub-Networks on Energy

Conservation from Occupancy Interventions in a Typical U.S. Commercial
Building.” American Society of Civil Engineers, 267–274.
Chen, J., and Ahn, C. (2014). “Assessing occupants’ energy load variation through
existing wireless network infrastructure in commercial and educational
buildings.” Energy and Buildings, 82, 540–549.
Clevenger, C., Haymaker, J., and Jalili, M. (2014). “Demonstrating the Impact of the
Occupant on Building Performance.” Journal of Computing in Civil
Engineering, 28(1), 99–102.
Gulbinas, R., Khosrowpour, A., and Taylor, J. (2015). “Segmentation and
Classification of Commercial Building Occupants by Energy-Use Efficiency
and Predictability.” IEEE Transactions on Smart Grid, PP(99), 1–1.
Gulbinas, R., and Taylor, J. E. (2014). “Effects of real-time eco-feedback and
organizational network dynamics on energy efficient behavior in commercial
buildings.” Energy and Buildings, 84, 493–500.
Kavulya, G., and Becerik-Gerber, B. (2012). “Understanding the Influence of
Occupant Behavior on Energy Consumption Patterns in Commercial
Buildings.” Computing in Civil Engineering (2012), 569–576.
Lassabe, F., Canalda, P., Chatonnay, P., Spies, F., and Charlet, D. (2006). “Refining
WiFi Indoor Positioning Renders Pertinent Deploying Location-Based
Multimedia Guide.” 20th International Conference on Advanced Information
Networking and Applications, 2006. AINA 2006, 126–132.
Meier, A. (2006). “Operating buildings during temporary electricity shortages.”
Energy and Buildings, Energy and Environment of Residential Buildings in
China, 38(11), 1296–1301.
Rieur, E., Alahmad, M. (2014). “On the Discourse of Energy as Material: Future
Feedback Technologies and Directions for Experiencing Energy,” IEEE
Trans. on Industrial Informatics, Volume: 10, Issue 1, Feb 2014, pages 742-
751
© ASCE
Staats, H., van Leeuwen, E., and Wit, A. (2000). “A longitudinal study of
informational interventions to save energy in an office building.” Journal of
Applied Behavior Analysis, 33(1), 101–104.
United Nations Environment Programme. (2007). “Buildings Can Play Key Role In
Combating Climate.”
U.S. Department of Energy. (2011). “Building Energy Data Book.”
U.S. Energy Information Administration. (2014). “Annual Energy Review.”
Zoha, A., Gluhak, A., Imran, M. A., and Rajasegarar, S. (2012). “Non-Intrusive Load
Monitoring Approaches for Disaggregated Energy Sensing: A Survey.”
Sensors, 12(12), 16838–16866.
© ASCE
Default Conditions: A Reason for Design to Integrate Human Factors
Arsalan Heydarian1; Joao P. Carneiro2; Evangelos Pantazis3; David Gerber4; and

Burcin Becerik-Gerber5
1,3
Ph.D. Student, Sonny Astani Department of Civil and Environmental Engineering,
University of Southern California, Los Angeles, CA. E-mail: heydaria@usc.edu;
epantazi@usc.edu
2
Graduate Student, Sonny Astani Department of Civil and Environmental Engineering,
University of Southern California, Los Angeles, CA. E-mail: jcarneir@usc.edu
4
Assistant Professor, School of Architecture and Sonny Astani Department of Civil and
Environmental Engineering, University of Southern California, Los Angeles, CA. E-mail:
dgerber@usc.edu
5
Associate Professor, Sonny Astani Department of Civil and Environmental Engineering,
University of Southern California, Los Angeles, CA. E-mail: becerik@usc.edu
Abstract
Increasing the use of natural light in commercial buildings could save up to 50

percent in energy consumption. Previous research has shown that there is a direct
connection between occupant behavior and the energy consumed in a building.
Experimental studies suggest occupants are less likely to adjust the shades if the
available lighting is within their preferred levels and enough to perform their daily
tasks. Therefore, in order to be able to increase daylighting in an office space, it is
important to understand the effect of default lighting settings on occupants’ rate of
lighting adjustments as well as their performance. In this paper, through the use of
immersive virtual environments, the authors have analyzed 64 participants’
interactions and performance in a virtual office space when put in a room with default
light setting set at either (1) only natural light available or (2) only artificial light
available. The preliminary results indicate that people are significantly more likely to
keep the lighting setting if the default has only natural light available than only
artificial light.
INTRODUCTION
Buildings account for roughly 45 percent of the energy consumption in the

United States (EPA 2013). In the past decade, there have been significant
improvements in technological advancements (e.g. automated lighting and shading
systems, smart thermal control systems and etc.), which have provided various
opportunities to reduce the total energy consumption in residential and commercial
buildings. However, technology alone does not guarantee low energy consumption
since most technologies are operated by occupants (Hong et al. 2015). Studies have
© ASCE
shown occupant behavior (actions and reactions of a person in response to external or

internal stimuli) could affect the energy consumption in buildings by a factor of three
(Fabi et al. 2013) and in some cases up to a factor of 10 (Sieminski 2013). It is
important to understand the driving factors that could impact occupant behavior in
interacting with different building systems and identify how these factors could result
in a reduction of building energy consumption.
Lighting systems are considered to be the second highest energy consumption
source in commercial buildings (following HVAC systems), accounting for
approximately 71 percent of the total electricity consumption in buildings in the U.S.
(EPA 2014). Research suggests that the use of daylight could significantly reduce the
total electricity consumption of lighting systems in commercial buildings (Van Den
Wymelenberg 2012). In buildings with poor design of daylight integration, studies
have observed an increase in energy consumption due to higher user interaction with
shading and artificial lighting systems (Dubois and Blomsterberg 2011). Additionally,
studies have identified positive effects on occupant mood, performance, and health
and well-being as other benefits of day lit environments (Galasiu and Veitch 2006).
Through a set of questionnaires responded to by 794 people working in two
commercial buildings in Japan, (Inoue et al. 1988) concluded that occupants adjust
the shades mainly due to visual comfort. In a similar study, through a survey of 113
participants, (Kawase 2000) reported that the two most common reasons for the
occupants to open the shades were (1) to increase daylighting in working spaces and
(2) maintain visual contact with the outside environment (75% and 62%,
respectively). Although visual contact and daylighting are important factors, previous
studies also indicate that occupants do not adjust shading frequently if the available
lighting is within occupants’ preferred levels and enough to perform daily activities
(Dubois and Blomsterberg 2011; O'Brien et al. 2013; Van Den Wymelenberg 2012).
Therefore, it is important to identify different ways to increase the use of daylighting
in order to decrease the electricity consumption while keeping the occupants satisfied.
In this work, we study the effect of default lighting conditions on occupants’
behavior and performance. As the first step into this work, we have chosen two
default lighting conditions that the participants can perform daily office related
activities in: (1) a room with all shades open and no artificial light on (maximum
natural light only) and (2) a room with all artificial lights on and all shades closed
(maximum artificial light only). Specifically, we explore the participants’ propensity
to keep or adjust the default lighting settings in order to better understand the effect of
such lighting conditions on participants’ behavior and performance when given a task
to perform (i.e., reading a passage and answering a set of questions). The paper
presents the research methodology, including the immersive virtual environment used
for data acquisition, as well as the results and analysis of participants’ decisions to
keep or change the default settings, and lastly a discussion on the proposed approach
and planned future work.
MOTIVATION TO USE IMMERSIVE VIRTUAL ENVIRONMENTS
Previous experimental studies have collected data through surveys or field

observations from office spaces and their occupants. Although office environments
© ASCE
provide the necessary related data on human behavior, these studies could be limited
in terms of their size and generalizability (e.g., small number of participants, one or
limited number of case studies, and etc.). More importantly, there is a possibility of
experimental noise that could affect the results and that cannot be controlled for (e.g.
changes in weather, daylighting availability, interior designs, location and orientation
of building(s) etc.). For instance, (Maniccia et al. 1999) mentions, in their study, the
difference in interior designs among different offices could have impacted the
shading and dimmer use among occupants of different buildings.
In order to reduce the experimental noise, collect accurate occupant related
information, and control different variables that could directly or indirectly affect
occupant behavior, we have utilized immersive virtual environments (IVEs) to create
virtual office spaces with realistic interior designs (e.g., tables, chairs) and lighting
features (e.g., shades, light switches, realistic luminance and illuminance) that a
physical office would have. In recent years, with the advents in fields of virtual and
augmented reality, there has been a significant amount of improvement in visual
quality of such environments, as well as their immersivity and interactivity options,
allowing users to feel fully immersed in the virtual environment. Previous studies
have shown that participants’ sense of presence is similar in IVEs and physical
environments (Adi and Roberts 2014). In our previous study, we benchmarked 150
participants’ performance and sense of presence between a physical office space and
a virtual office space (Heydarian et al. 2015); we concluded that participants perform
similarly between the two environments and have a strong sense of presence within
IVEs. Another important advantage of IVEs is that the experimenter is less salient to
the participant (as participants cannot see the experimenter), facilitating behavior that
is more natural. Therefore, in this study, we utilized IVEs to facilitate our
explorations.
METHODOLOGY
In order to develop a better understanding of occupant’s propensity to keep or

adjust default lighting conditions, 64 participants were recruited to perform a set of
activities in a virtual office. Through observations of the participants’ interactions
within the virtual office (e.g., keeping or changing the default settings) and noting
what lighting features were kept or adjusted (e.g. number of light bulbs that were
turned on/off and number of shadings that were opened/closed), the authors analyzed
how the initial lighting settings could influence occupant’s lighting choices.
Additionally, the authors assessed how individual’s environmental values affected
their propensity to change the default conditions as well as their final lighting choice.
Experiment Design. To evaluate our research objective, a virtual office (conference

room) similar to an actual office was designed. Although there are many design
alternatives that could affect the amount of available lighting in an office space (e.g.
types and sizes of windows, type of light bulbs, office furniture, and etc.), the authors
chose generic window types and reduced the number of furniture in order to reduce
any reflective surfaces that might affect the participants’ decisions during the
experiment (Figure 1). The modeled office space was 150 square meters in size, and
© ASCE
hadd three windo ows (includiing manually y controlled shades) andd 12 light fixxtures (three
fluoorescent ligh ht bulbs on each
e fixture). The light sswitches in tthe room weere designed
in a way that the t participaant could adjjust the artifficial lightinng to have oone, two, or
threee light bulb bs on each fixture.
f Figure 1 shows the renderedd model of tthe test bed
envvironment.
The parrticipants were
w asked tot read a shhort passagee and answ wer a set off
commprehension n questions based on what w they hhave read. The particiipants were
randomly assig gned either to a room with the deefault settingg of only nnatural light
avaailable (all sh hades open and
a no artifiicial lights oon) or only aartificial lighhts available
(alll artificial liights on and d no shadess open). Thee authors chhose these ttwo default
lighhting settings as they aree the polar opposites
o andd thus allow the authors to examine
the extent to wh hich a defau
ult lighting setting
s influeence a particcipant’s lightting choice.
Theey were instrructed that th heir task is to
t read the pprovided passage and thaat they have
the option of either
e keepiing the defaault light settting or adjusting it to their most
preferred settin ng. The partiicipants had the option tto open/closse each set oof shades to
incrrease/decreaase the availaability of naatural light aand turn the light switchhes on/off to
conntrol the arrtificial ligh ht levels in n the room. Followingg the experriment, the
parrticipants com mpleted a qu uestionnaire assessing thheir environm mental valuees.
Figure
F 1 – Virtual
V Officce Space
Mo odel and Ap pparatus. To o be able to create a reaalistic and innteractive virtual model
andd have the participants
p to be fully immersed inn them, a 33D model off the office
spaace was first designed in n Revit© 2015. The Reviit model onlyy consisted of the basic
geoometry of th he room, inccluding wallls, generic w windows, thee ceiling, annd the light
fixttures. The Revit
R model was then im mported to 3dds Max© to give it a moore realistic
look by adding g furniture, materials,
m teexture, shaddows, reflecttions, and arrtificial and
natuural lighting
g. Once all the
t necessary y componennts to make the room loook realistic
©
werre added in 3ds Max , 32 3 different renderings ((render-to-teexture) were created for
eveery lighting combination n that the paarticipants ccould have iin the IVE. In order to
enssure the lighting settingss in 3ds Max x© were reallistic and reppresentative of real-life
valuues, 32 diffeerent light maps
m were geenerated usinng Ladybugg and Honeyybee plugins
(op
pen source plugins for Grasshopper3
G 3D – Rhino plugin (Rouudsari et al. 2013)) and
© ASCE
the lux values around the room were calculated. It is importtant to note in order to
keeep everything constant among
a the participants,
p the modeledd office wass located in
Loss Angles an nd the simu ulations weere performeed at 2:00 PM, June 22nd; both
connditions’ avaailable lux values
v were according tto the standaard’s for intterior office
spaace with lux values approximately ranging
r fromm 500 to 10000 lux throoughout the
dayy.
32 models were then exported as a FBX files from 3ds M Max© and were imported
© ©
in Unity
U gamee engine as different
d scen
nes. The Unnity game eengine was uused for two
purrposes: (1) to
o program in nteractive opptions for eacch scene andd (2) to connnect the IVE
equuipment to thhe 3D models. The interractive optioons were proogrammed inn a way that
wou uld allow thhe participaants to havee a more re alistic interaaction with the virtual
models (e.g., being able to turn light bulbs
b on or ooff by standiing in front of the light
swiitch or opening and closing
c the shades by standing inn front of a window,
animmations thaat would show the sh hades openning or cloosing based on users’
inteeraction). Th
he equipmen nt that were used
u to provvide a fully iimmersive eenvironment
inclluded an Occulus DK2 Head-Moun nted Displayy (HMD), ann Xbox-3600 controller,
andd a positional tracker thaat would track the particiipants’ head and neck movements.
Exp perimental Procedure. 64 participants, betweeen ages 18 tto 32, were recruited (a
powwer analysis confirmed thet sample size
s was suffficient to dettect significaant effect at
a power
p of 0.990). The parrticipants were
w all undeergraduate oor graduate students in
diffferent fields of study (e.g. engineeriing, architeccture, life-scciences, art, and etc.) at
the University of Southern n California. Prior to thee experimennt, the participants were
given a brief explanation
e of
o the experriment withoout revealinng any inforrmation that
cou
uld affect their judgmen nt in their deecisions durring the experiment. Following the
exp
planation, th hey were askked to read and sign ann IRB (Instittutional Revview Board)
app
proved conseent form.
Figuree 2 – A partiicipant interracting and n

navigating w
within the IV
VE
Oncce the particcipants sign

ned the consent form, thhey were givven a brief training on
ng the IVE equipment (e.g., learn
usin ning how too use differeent keys onn the Xbox
con
ntroller). Theey were then
n instructed that they woould have too read a passsage located
on the table in the middle of
o the room and they hadd the option of keeping the lighting
in the
t room or adjusting itt to their preeferred settiings by interracting withh the shades
© ASCE
and/or the light switch. The participants were randomly divided into two groups with
different default light settings. They were asked to place the HMD on their head and
walk inside the virtual office. Once they entered the room, at first, they were
instructed to walk around the space and get a feeling of the environment (Figure 2). If
they were in a default light setting with “only natural light available”, they only had
the shades open and all the artificial lights were off, and if they started with “only
artificial lights” available, they had all the shadings closed and maximum artificial
light available.
If they chose to change the default setting, the setting they set as their
“preferred” setting was recorded. Once comfortable with the lighting setting (either
the default setting or the preferred setting), they were then asked to read over the
provided passage on the table. The amount of time that took the participants to read
the passage was recorded. Upon completion of the reading task, they were asked to
remove the HMD and complete a comprehension test based on what they had read.
Following that, they were asked to complete an environmental value questionnaire.
RESULTS AND ANALYSIS
We first examined how the lighting default settings (“all natural lights” and
“all artificial lights) differentially influenced participant’s decision to keep or change
the lighting setting. There was a statistically significant difference in participants’
propensity to keep the default light setting between the two lighting conditions, such
that those with the “all natural lights” default setting were significantly more likely to
keep the default setting than those with the “all artificial lights” default setting, χ2 (1)
= 9.94, p = .002. 74% of the participants with the “all natural lights” default setting
kept the initial lighting setting whereas only 29% of the participants in “all artificial
lights” default kept the initial light setting.
Participants were marginally significantly more likely to keep the initial
lighting setting than at chance (50% chance of keeping and 50% chance of changing)
when they had “all natural lights” as the lighting default, 74% vs. 50%, χ2 (1) = 2.89,
p = .09. However, participants were not significantly more likely to keep the initial
lighting setting than at chance when they had “all artificial lights” as the default, 26%
vs. 50%, χ2 (1) = 2.33, p = .17.
Participants with the “all-natural light” default setting were significantly more
likely to have all of the blinds open as their final choice than those with the “all
artificial light” default setting (94% vs. 32% χ2 (1) = 25.69, p < .0001). Figure 3
shows the distribution of the final lighting choices for each of the default light
settings.
© ASCE
Figure 3 – Distribution of final ch

hoices for eaach of the d
default light setting
Overall, the samplle of particiipants in booth the “alll natural” ddefault light
conndition and “all artificial” defau ult light coondition weere similarlyy high on
envvironmental--friendliness according to o the adminiistered questtionnaire, M = 6.13 and
M= = 6.86 (out of a maximum m score of 10), respectivvely. Particippants who wwere high on
envvironmental friendliness were signifi ficantly moree likely to chhange the innitial setting
oveerall, β = -.35, p = .032. However, th here was no significant eeffect of envvironmental
frieendliness on the final cho
oice.
Particip
pants in the “all
“ natural light” defauult setting annswered com mprehension
testt questions correctly marginally
m more
m significcantly than those in thee “artificial
lighht” default condition, β = 12.32, p = .061. Theree were no siignificant diffferences in
read ding speed between
b the two settingss.
ONCLUSION, LIMITA
CO ATIONS AN
ND FUTURE
E WORK
These results
r sugg
gest that deefault settings can havve different effects on
inflluencing parrticipants to keep the in nitial lightingg setting. Lighting defaault settings
with all naturall lights weree more likely y to be kept by participaants than at chance and
more likely to be b kept than n artificial lig
ght default ssettings. As a result, parrticipants in
with the all-nattural light deefault setting g were signiificantly morre likely to hhave all the
shaades open in n their final choice than those in the all artificiial light defaault setting.
Theese participaants also perrformed bettter on the coomprehensioon questionss than those
in the artificial light defaultd settiing. Additiionally, thee authors found that
envvironmentally friendly individuals were w more llikely to chhange the innitial setting
oveerall. These results have implicattions for arrchitects, enngineers, annd building
man nagers. As natural
n lightt use has beeen shown too be an effeective meanns to reduce
eneergy, office rooms shou uld be design ned with wiindows and managers oof buildings
sho
ould have thee blinds opeened as a default to encoourage naturral light use rather than
artiificial light use.
u
This reesearch brid dges the gap p between the human behavior rresearch on
occcupant behav vior and eng gineering/arcchitectural w work on buildding design and energy
© ASCE
efficiency. The work contributes to the literature in the architecture/engineering

domain by demonstrating an effective way to design and/or operate a building in
order to reduce energy consumption. It contributes to the literature in human behavior
by revealing how (1) default settings influence lighting choices in an office
environment and (2) how they can have different effects on occupants’ lighting
choice depending on what the particular default setting is (e.g., all natural lights vs.
all artificial lights).
The results presented in this paper are the first steps towards better
understanding the effect of default lighting settings on occupants’ rate of lighting
adjustments, as well as their behavior. As part of on-going work, we are exploring
how different combinations of artificial light and natural light may affect an
individual’s propensity to keep or change the default condition. Additionally,
different architectural design options (e.g., spatial configurations, room geometries,
proportions, orientations and shapes, type and size of windows, materials, etc.) could
affect the available lighting levels in an office environment. In our future work, we
will evaluate the effects of various designs and design features on users’ decisions
and energy relevant behavior.
These findings open the doors to a wide range of future research questions
that we are in currently exploring. For example, in this research the differential effect
of default settings was shown between all artificial light condition and all natural light
condition. We are interested in integrating the observed results from this study in the
design phase of buildings. To do this, we are interested in creating systems to develop
new alternative designs and improve models that reduce the energy consumption of
buildings while meeting end-user needs and preferences.
ACKNOWLEDGMENT
This material is based upon work supported by the National Science

Foundation under Grant No.1231001. Any opinions, findings, and conclusions or
recommendations expressed in this material are those of the author(s) and do not
necessarily reflect the views of the National Science Foundation. Special thanks to all
the participants and researchers that contributed to this project; specifically Saba
Khashe for her contribution on helping with preparing and running the experiments.
REFERENCES
Adi, M., and Roberts, D. (2014). "Using Virtual Environments to Test the Effects of
Lifelike Architecture on People." Technologies of Inclusive Well-Being, A. L.
Brooks, S. Brahnam, and L. C. Jain, eds., 261-285.
Dubois, M.-C., and Blomsterberg, Å. (2011). "Energy saving potential and strategies for
electric lighting in future North European, low energy office buildings: A
literature review." Energy and Buildings, 43(10), 2572-2582.
EPA (2013). "National Awarness of Energy Star ",
<http://www.energystar.gov/sites/default/uploads/about/old/files/2013%2020CEE
%2020Report_2508%2020compliant.pdf>.
EPA (2014). "Energy Star, Building and Plants."
<http://www.energystar.gov/buildings?s=mega>.
© ASCE
Fabi, V., Andersen, R. V., Corgnati, S. P., and Olesen, B. W. "A methodology for
modelling energy-related human behaviour: Application to window opening
behaviour in residential buildings." Proc., Building Simulation, 415-427.
Galasiu, A. D., and Veitch, J. A. (2006). "Occupant preferences and satisfaction with the
luminous environment and control systems in daylit offices: a literature review."
Energy and Buildings, 38(7), 728-742.
Heydarian, A., Carneiro, J. P., Gerber, D., Becerik-Gerber, B., Hayes, T., and Wood, W.
(2015). "Immersive virtual environments versus physical built environments: A
benchmarking study for building design and user-built environment
explorations." Automation in Construction, 54(0), 116-126.
Hong, T., D'Oca, S., Turner, W. J. N., and Taylor-Lange, S. C. (2015). "An ontology to
represent energy-related occupant behavior in buildings. Part I: Introduction to
the DNAs framework." Building and Environment(0).
Inoue, T., Kawase, T., Ibamoto, T., Takakusa, S., and Matsuo, Y. (1988). "The
development of an optimal control system for window shading devices based on
investigations in office buildings." ASHRAE transactions, 94, 1034-1049.
Kawase, T. (2000). "Office Worker Preferences of Exterior Shading Devices: a Pilot
Study." Eurosun 2000, Copenhagen, Denmark.
Maniccia, D., Rutledge, B., Rea, M. S., and Morrow, W. (1999). "Occupant use of
manual lighting controls in private offices." Journal of the Illuminating
Engineering Society, 28(2), 42-56.
O'Brien, W., Kapsis, K., and Athienitis, A. K. (2013). "Manually-operated window shade
patterns in office buildings: A critical review." Building and Environment, 60(0),
319-338.
Roudsari, M. S., Pak, M., and Smith, A. "ladybug: A Parametric Environmental Plugin
For Grasshopper To Help Designers Create An Environmentally-Conscious
Design." Proc., 13th conference of international building performance
association, chambery, france, 26-28.
Sieminski, A. (2013). "International energy outlook 2013." US Energy Information
Administration (EIA) Report Number: DOE/EIA-0484.
Van Den Wymelenberg, K. (2012). "Patterns of occupant interaction with window blinds:
A literature review." Energy and Buildings, 51, 165-176.
© ASCE
Assessing Energy Strategies in Active Buildings Considering Human Behaviour
Ayesha Kashif1,2; Stephane Ploix1; and Julie Dugdale2

1
G-SCOP Lab/Grenoble Institute of Technology, 46 Avenue Felix Viallet, 38031
Grenoble, France. E-mail: ayesha.kashif@g-scop.inpg.fr; stephane.ploix@grenoble-
inp.fr
2
LIG Lab/University of Grenoble, 110, Av de la Chimie, 38400, Saint Martin
d'Héres, France. E-mail: julie.dugdale@imag.fr
Abstract
In the recent years, surveys and studies have established the importance of
occupant's behaviour on energy consumption in buildings. Therefore, inclusion of
inhabitants' behaviours is compulsory for the assessment of building energy
management system's (BEMS) strategies, which highly depends on human
behaviour. The purpose of modelling the inhabitants’ behaviour is to see how their
choices and control of household appliances can impact the energy consumption. In
this paper, a co-simulation approach is presented where the inhabitants' behaviours
are co-simulated with the SIMBAD-MOZART thermal model of a reference house
and BEMS. The realization of all the different kinds of inhabitant behaviours into
energy co-simulations will help to improve the smart grid technology and hence
provide inhabitants with better services to save energy and cost while maintaining
their comfort levels.
INTRODUCTION
The advancements in the electric grid technology have led to the concept of a
smart grid that uses the information technology to communicate with the suppliers
and customers about their energy supply and demand needs. The smart grid helps in
improving energy efficiency and sustainability of its production and distribution. The
information that can be provided to the inhabitants consists of availability of energy,
tariff details and energy consumption by different household appliances etc. After
receiving all the different information from the smart grid, the inhabitants must be
intelligent enough to interpret all this information so that they can save energy while
maintaining their comfort. This requires a high cognitive workload to make decisions
about energy management, and the results depend on how intelligently the
information is handled and acted upon.
The intelligent systems called Building Energy Management System (BEMS)

are under development (Doukas et al., 2007). They control the environmental
conditions inside the house such that its less costly and more comfortable for the
© ASCE
inhabitants. Thhe inhabitannts can alsoo communiccate with thhe BEMS and a can exppress
theiir comfort needs,
n occuppancy planss etc. and caan also ask for
f advice.
In ordeer to assess and evaluaate the diffeerent strateggies that aree developedd by
the BEMS, it i is imporrtant to innclude the inhabitantss’ reactive and dynaamic
inteeractions with
w their ennvironment in buildingg energy siimulations. It will helpp to
anaalyze the coontrol of diifferent behhaviours oveer the envirronment annd the resullting
imppact on enerrgy consum mption patterrns. Similarrly, the rolee of BEMS in the preseence
of these reactiive behavioours will be more chaallenging annd will leaad to improoved
funnctionality and
a energyy efficient decision
d m
making. Thee BEMS ussed in the co-
sim
mulation callled G-HomeeTech (Ha et al., 2012) has been developed
d a G-SCOP and
at
commmercializeed by Vesta System (VeestaEnergy,, 2011).
INH
HABITANTS' BEHA
AVIOUR MODEL
M
Figgure 1. H-B mic behaviour represen

BDI dynam ntation moodel
In this paper, diffe

ferent elemeents that connstitute the inhabitantss' behaviourr for
eneergy management are combined
c t build a global
to g H-BD DI model (F Figure 1). This
T
model is based on the BDI B (belief, desire, intention) archhitecture (G Georgeff et al.,
19999). Howevver, the nottion of hom meostasis iss introduced to capturre the physsical
behhaviour of human
h bodyy. Figure 1 shows thee cycle of inhabitants' behaviour that
starrts with peerception of o the env vironment, passes thro ough the instinctive
i and
coggnitive phasses and endds up with actions back on the environmennt. The outside
envvironment inncludes the location, physical buillding models, the objects, appliannces,
andd other agen nts etc. Alll these envvironmental elements are a then peerceived by the
ageent. Upon thet perceptiion the ageent will traanslate these elements as its beliiefs,
shoown by the "Beliefs" part of the cyycle. In thee model in figuref 1, hoowever, anoother
conncept is introoduced in addition
a to beliefs
b that rrelates to thhe internal physical
p statte of
an inhabitant e.g. hungerr level based on metaabolism. Baased on thee beliefs abbout
hommeostasis anda the ouutside envirronment thhe agent can c have certain
c desiires,
© ASCE
howwever, due to
t the externnal environmental consstraints only
y one of theem is conveerted
to tthe agent's intention. Finally,
F bassed on the intention thhe agent peerforms cerrtain
actiions on the environmennt.
CO
O-SIMULA
ATION ENV
VIRONME
ENT
Reactiv ve and deliberative behaviour

b m
models of inhabitants
i can easilyy be
impplemented withw a multti-agent appproach (Kasshif et al., 2013a)2 and
d (Kashif ett al.,
20113b). The same multti-agent appproach is uused to coo-simulate the t inhabittants
dynnamic behav viour with BEMS
B that will help too analyse thhe strategies developedd by
BEMS in the presence of o occupannts in the house.
h Thiss section prresents the co-
mulation off inhabitantts' behaviouur with thhe thermal model, SIIMBAD, of
sim o a
refeerence buildding, MOZ ZART (Noel, 2008) annd the Buildding Energy y Managem ment
Sysstem (BEM MS) G-HomeeTech. Thee objective is to analy yse the impaact of buildding
eneergy management systeem to save energy in the presencce of inhabbitants' reacctive
andd dynamic decision
d maaking behavviour on hoousehold apppliances. A comparisoon is
alsoo made to analyse the impact of o different behaviourss (Eco, Noon Eco) on the
eneergy consum mption and thermal coomfort leveels with andd without thhe presencee of
BEMS. The notionn of coomfort in thhe inhabitaants is introoduced usinng the Fangger's
commfort modell (Fanger, 1973).
The theermal modeel for this house
h calledd the SIMBA AD-MOZA ART model was
built in Matlabb/Simulink by CSTB (Centre( Scientifique ett Techniquee du Batimeent).
MBAD-MO
SIM OZART calcculates the temperature
t e in each zoone by takinng into accoount
varrious input variables. Some of th he most im mportant vaariables, shoown inside the
yelllow rectanggle in figuree 2, include the power of all the diifferent app pliances present
in tthe zone, thhe position of the blind ds e.g. openn/closed, nuumber of occcupants inn the
zonne, respiratiion flow raate, weatheer data, artiificial lightting, and ventilation.
v The
imppact of wiindow statees (openedd/closed) iss also takeen into acccount throough
venntilation, i.e. the air maass flow betw
ween the inside and ouutside of thee building.
Figure 2. SIMBA RT thermal model

AD-MOZAR
© ASCE
The Brrahms-SIMB BAD-G-HoomeTech coo-simulationn environmeent is shown in

figuure 3. Thee Brahms-B BEMS-Interrface moduule providees the interrconnectionn of
SIMMBAD theermal modeel with booth the BE EMS and the Brahm ms multi-aggent
sim
mulation envvironment (S Sierhuis, 20008). The innput that goes to this module
m from
m the
SIMMBAD therrmal modeel is the aiir temperatture, mean radiant temperature and
hummidity. Othher inputs include thhe electric power of appliancess, the setppoint
tem
mperature annd the appliiance modee (on/off). The T BEMS will w use theese variablees to
commpute the en nergy plan and to conttrol the apppliances. Coonversely, inn Brahms thhese
varriables are perceived
p byy the agentss, who furthher take certtain actions to control their
t
therrmal enviroonment.
The ouutput from m this interrface moduule either comes c from
m the Brahhms
mulation en
sim nvironment or the BEMS. B Thee output fromf Brahm ms simulaation
envvironment consists
c off occupancy data in each room m in the house
h and the
stattus/modes (on/off,
( opeen/closed) of
o all househhold appliannces or objects. Similaarly,
the output from m the BEMS S consists of
o the setpoiints and apppliance moddes.
The SIIMBAD-MO OZART-Thhermal-Moddel module in figure 3 continuouusly
perrceives the values com ming fromm either thee BEMS or the Brahhms simulaation
envvironment anda calculattes the new temperaturre at each simulation step. s MOZA ART
houuse is furthher used forr developinng a scenariio of inhabbitants' pressence and their t
actiivities. Thee purpose of modellingg the inhabbitants' behaaviour is too see how their t
chooices and co ontrol of houusehold app pliances cann impact thee energy connsumption.
Figgure 3. Co--simulation
n environm
ment
O-SIMULA
CO ATION SCE
ENARIO AND
A RESU
ULTS
A scen nario of a 2 person fammily, husbannd and wife,, has been implemente
i ed in
Braahms. The husband is an "Eco agent",a wheereas the wife
w is a "N Non-eco" aggent.
Aftter spendingg their day at
a work, thee agents com me back hoome in the evening.
e Ass the
houuse is relativvely warmeer than outsiide, the ageents will perrceive it to be comfortaable
for a while. However,
H fter a short period theyy will start perceiving that the acctual
aft
© ASCE
tem
mperature iss very low w. The com mfort/discommfort of ann agent is based on the
hommeostasis which
w furtheer depends on the percceived PMV V (Predictivve Mean Vote)
V
valuues in this scenario.
s Ass soon as an
n agent startts to feel waarm (PMV 1 to 2), it taakes
somme action too be comforrtable again.. However, if it does no ot take any action or iff the
actiion does noot result thhe agent beeing comforrtable againn it will sttart feeling hot
(PMMV 2 to 3) or too hot (above 3). Similarly, as a soon as the t agents start
s perceivving
neggative PMV V value, theyy increase thhe temperatuure setpointt to be warm mth.
Since their
t percepption of com mfort does not
n solely depend
d on thhe temperatture,
butt also on othher factors, i.e. what acctivity they are involved in, what clothes
c theyy are
weaaring etc. The
T time att which theyy feel com mfortable varries. As soon as an aggent
starrts to feel warm
w it willl take an acction to be comfortablle again. Thhe EcoHusbband
ageent would prefer
p to decrease
d thee temperatuure by rem moving extraa clothing and
turnning off thee heater wheereas the NoonEcoWife agent woulld like to oppen the winddow
to quickly
q beccome comfoortable, witthout caringg that the heater
h that is
i still workking
andd that it is wearing
w tooo many clotthes. The innformation about
a the control overr the
apppliance/objeect is sent too the SIMB BAD thermaal model, wherew the neew temperaature
for the room iss calculatedd and sent back to Brahhms. Based upon the neew temperaature
the PMV valuees for all thee agents aree again calcuulated.
Figgure 4. Braahms simullation: inhaabitant's beehaviour an

nd BEMS'ss control ovver
e
environmen nt
Figure 4 shows a situation where w the BEMS

B turnned on the heater an hour
h
beffore the agennts enter thee living rooom, shown bby the yelloow colouredd workframees at
arouund 15:00 in EnergyM Manager's sppace. The reason
r for this
t is that energy tarifff is
low
w at this houur of this daay. The EcooHusband agenta expreessed discommfort to BEEMS
andd BEMS adjjusted the hheater to a new n value. This comm munication is shown byy the
yelllow coloureed workfram mes at arouund 16:20 inn EcoHusband and EnnergyManagger's
worrkspace. Th his time hoowever, whhen the BE EMS increaased the teemperature and
EcooHusband agent
a startedd feeling com mfortable, it
i removed its sweater, shown by "put
"
off sweater" tool tip at this workkframe arouund 17:30. This caussed him too be
© ASCE
unccomfortablee with the seetpoint adjusted by the BEMS and d it did not communicat
c te to
the BEMS. It rather itsellf increasedd the setpoiint to a higgher value and a put on the
sweeater. This is
i shown byy the "Adjust Heater Seetpoint" toool tip and "pput on sweaater"
tool tips on thheis workfraame in EcoH Husband's sspace at aroound 17:40. The blue line
going from thiis workfram me to the woorkframe in Livingroom mHeater's sppace shows that
the EchoHusbaand agent directly
d conttrolled the heater
h withoout any intervention byy the
eneergy manageer. These acctions helpeed the agent to become comfortablle shown byy the
yelllow coloured upward arrow shoowing the juump from one thermaal conditionn to
anoother in figgure 5c at around
a 18:000. The temmperature further
f wennt up to 266°C,
shoown in figurre 5b at aroound 19:30.. Now againn it starts feeling
fe warmm and turnss off
the heater. At this
t point wwhen the temmperature sttarts decreaasing, the BE EMS interruupts
the agents' deccisions and does not lett the temperature fall below
b 23°C
C by controllling
the heating system.
s Thhe state of the heatter is shown in fig gure 5a unnder
"LivingroomH Heater State"" where thee signal first goes to zeero and thenn to one duue to
BEMS interrupption. This is shown by b the "Set Temperaturre Intelligen ntly" tool tip in
EneergyManageer's worksppace at arouund 19:30. ThusT the EcoHusband d agent remmains
commfortable with
w the decision taken by b the BEM MS shown byy the green curve in figgure
5c between 200:00 and 233:00. The teemperature when contrrolled by thhe BEMS, also
helpps NonEcooWife agentt to remainn in the sliightly cool to comforttable condiition
rathher than being cool orr cold (figure 5c, 5d). This is sho own by the light blue and
whiite workfram mes in NonnEcoWife's space in figgure 4 and byb the light blue and grreen
currve in figuree 5d betweenn 20:00 andd 23:00.
(b) T
Temperaturre controlledd by agent and
a
BEMS
(aa) Agent co
ontrols heateer with BEM
MS
(c) PMV perceeived by EcoHusband agent

a (d) PMV
P perceiv
ved by NonnEcoWife aggent
Figgure 5. Statte of the appliance/objject, tempeerature, and
d PMV perrceived durring
simula
ation with B BEMS
© ASCE
CO
OST AND COMFORT
C T ANALYS
SIS
In this section, an analysis off the cost-coomfort tradeeoff for the situations with
w
andd without th he BEMS iss given. To quantify thhe comfort of o agents, thhe PMV vaalues
obttained after the simulattion runs arre summed up for diffferent PMV levels (figuures
6a and 6b). Siince EcoHuusband agennt is not onlly concerneed by the coomfort but also
the energy savvings and inn this efforrt it remainss less comffortable thann NonEcoW Wife
ageent (figure 6a).
6 Mostly, it remainss in slightlyy cool or slightly warm m due to havving
morre interactions with thhe heater to control the t temperaature. NonE EcoWife aggent,
howwever, remaains more comfortable
c e than EcoH Husband ageent, as it is not concerrned
aboout energy savings
s andd wants to achieve
a commfort at any cost. Figurre 6b showss the
therrmal comfoort durationss of agents with the innclusion of a BEMS inn the system m. In
thiss case, the divergence
d o agents' coomfort levells is reduced and they converge
of c too the
commfortable zoone. Also, the
t agents remain
r commfortable forr a longer tiime durationn as
commpared to before
b i.e. without
w BEM MS. In this case EcoH Husband ageent's comforrt is
bettter than NoonEcoWife agent. The improvemeent in the co omfort is duue to the beetter
deccisions takenn by the BE EMS based on the knoowledge thaat the BEMS S has aboutt the
inteernal and external ennvironmenttal conditioons, weathher forecastts, inhabitaant's
commfort and seelf learning algorithms..
((a) Agents' thermal

t com
mfort withouut (b) Agents' thermal comfoort with BEM
MS
BEMS
Figure 6. Comfortt of agents:: with and wwithout thee control off BEMS
Figure 7 shows the t power consumptioon of the electric heeater while the
envvironment is he BEMS. The
i controlleed by differrent agentss with and without th
highest power consumed is due to th he behaviouur of NonEccoWife ageent since it tries
t
to aachieve com
mfort by oppening and closing thee window. This
T assessm
ment of BE EMS
wheen co-simu ulated with building syystem and iinhabitants shows thatt the BEMS is
cappable of noot only savving the inhhabitants frrom cognitiive workloaad but alsoo of
prooviding them
m with betteer comfort and
a energy savings.
s
© ASCE
Figgure 7. Energy consum

med duringg control ovver environ
nment by diifferent ageents
with
h/without BEMS
CO
ONCLUSIO
ON
The beehaviour model in thee co-simulaator generattes the proffiles which are
random and dy ynamic. Ass soon as thhe environm mental variaables chang ge, they chaange
ageents’ beliefss and the system reacts in a differeent way thann before. Thhe introducction
of iinhabitants’ reasoning processes
p toowards their actions on n the physiccal environm
ment
will give energy simulatiion tools more
m realism
m. The reacttions to theese grid signals
couuld further be
b diverse and
a complex x dependingg on differen nt types of inhabitants e.g.
bassed on theeir family compositioon, role inn the familly, econom mic conditions,
knoowledge andd concerns about enerrgy problem m. The realiization of alla the diffeerent
kinds of inhabbitant behavviours into energy co--simulationss with the smart s grid will
helpp to improvve the smarrt grid techhnology andd hence pro ovide the innhabitants with
w
bettter services to save eneergy and cosst while maaintaining thheir comfortt levels.
RE
EFERENCE
ES
Douukas, H., Paatlitzianas, K.D.,

K Latropoulos, K., Psarras, J. (2007) Intelligent
buildiing energy managemen
m nt system ussing rule setts, Building and
Envirronment vol. 42(10) p. 3562–9
3
Fannger, P. (19773) Assessm ment of manns thermal comfort
c in practice.
p Briitish Journaal of
Industrial Mediciine 30, 313--324.
Geoorgeff, M., Pell, B., Poollack, M., Tambe,
T M., and Woolddridge, M. (1 1999) The
belieff-desire-inteention modeel of agencyy. In Proceed dings of 5thh Internationnal
Workkshop on Inttelligent Aggents: Agentt Theories, Architectur
A res, and
Languuages (Heiddelberg Germ many), Spriinger-Verlaag, pp. 1 - 10 0.
Ha,, D.L., Joummaa, H., Plooix, S., Jacomino, M. (22012) “An optimal
o appproach for
electrrical manageement probllem in dwelllings,” Eneergy and Buuildings, voll.
45, noo. 0, p. 1–144
Kasshif, A., Plooix, S., Duggdale, J., andd Le, X. H. B. (2013a) Simulatingg the dynamics
of occcupant behaaviour for poower managgement in reesidential buildings.
Energgy and Builddings 56, 85 5-93.
© ASCE
Kashif, A., Dugdale, J., and Ploix, S. (2013b) Simulating occupants’ behaviour for
energy waste reduction in dwellings: A multi agent methodology. Advances
in Complex Systems 16, 37.
Noel, J. (2008) Cas d’exemple codyba a partir de la typologie cstb des batiments,
http://www.jnlog.com/pdf/typologie_cstb.pdf >(July. 7, 2015)
Sierhuis, M., Clancey, W., and Van Hoof, R. (2007) Brahms - a multiagent modeling
environment for simulating work practice in organizations. International
Journal of Simulation and Process Modelling 3(3), 134-152.
VestaEnergy, (2011) Solutions logicielles d’energy management dynamique,
http://www.asprom.com/jei/vesta.pdf >(July. 7, 2015)
© ASCE
Estimating Occupancy in an Office Setting
Manar Amayri1; Stephane Ploix2; and Sanghamitra Bandyopadhyay3

1,2,3
ASCE Conference 2015 G-SCOP Laboratory/Grenoble Institute of
Technology, 1, 46 Avenue Felix Viallet, 38031 Grenoble, France.
E-mail: Manar.Amayri@grenoble-inp.fr
Abstract
A general approach is proposed to estimate the number of occupants in a

zone using different kinds of measurements such as motion detection, power
consumption or CO2 concentration. The proposed approach is inspired from
machine learning. It starts by determining among different measurements those
that are the most useful by calculating the information gains. Then, an estimation
algorithm is proposed. It relies on a C4.5 learning algorithm that yields human
readable decision trees using measurements to estimate the number of occupants.
It has been applied to an office setting.
INTRODUCTION
Recently, research about building turns to focus on occupant behaviors.

Most of these works deal with the design stage: the target is to represent the
diversity of occupant behaviors in order to guarantee minimal measured
performances. Most of the approaches uses statistics about human behaviors
(Rouletetal.,1991; Pageetal.,2007; Robinson and Haldi, 2009). (Kashif et al.,
2013) emphasized that inhabitants’ detailed reactive and deliberative behaviors
must also be taken into account and proposed a co-simulation methodology to
find the impact of certain actions on energy consumption.
Nevertheless, human behavior is not only interesting during the design
step, but also during operation. It is indeed useful for diagnostic analyses to
discriminate human misbehaviors from building system performance, but also for
energy management where strategies depend on human activities and, in
particular, on the number of occupants in a zone. Unfortunately, the number of
occupants is not easy to measure. This paper tackles this issue. It proposes an
occupancy estimator combining different measurements such as CO2
concentration, motion detection, power consumption because only one
measurement proved to be not reliable enough to estimate the number of
occupants. For instance, CO2 concentration may be useful but in some
configurations, when a window is opened for instance, estimations become
unreliable. Motion detection and power consumptions depend on occupant
activities. However, altogether, these measurements can be combined to get a
more reliable estimator.
© ASCE
STATE OF THE ART
Similar work for finding occupancy has been already tackled and various
methods have been investigated. The methods vary from basic single feature
classifiers that distinguish among two classes Presence and Absence to multi-
sensor, multi-feature models. A primary approach, which is prevalent in many
commercial buildings is to use passive infrared (PIR) sensors for occupancy.
However, motion detectors fail to detect presence when occupants remain
relatively still, which is quite common during activities like working on a
computer, or regular desk work. Furthermore, drifts of warm or cold air on objects
can be interpreted as motion leading to false positive detections. This makes the
use of only PIRs for occupancy counting purpose less attractive. Conjunction of
PIRs with other sensors can be useful as discussed in (Agarwal et al. 2010) who
makes use of motion sensors and magnetic reed switches for occupancy detection
to increase efficiency in HVAC systems of smart buildings, which is quite simple
and non-intrusive. Apart from motion, acoustic sensors (Padmanabh et al., 2009)
may be utilized. However, audio from the environment can easily fool such
sensors, and with no support from other sensors it can report many false positive
detections. In the same way, other sensors like video cameras (Erickson et al.,
2011; Milenkovic and Amft, 2013b), which utilize the huge advances in the field
of computer vision and the ever increasing computational capabilities, RFID tags
(Philipose et al., 2004) installed on id cards, sonar sensors (Milenkovic and Amft,
2013a) plugged on monitors to identify presence of a person on the computer,
have been used and have proved to be much better at solving the problem of
occupancy count, yet cannot be employed in most office buildings for reasons like
privacy and cost concerns. Pressure sensors and PIRs has been discussed in
(Nguyen and Aiello, 2012) to determine presence/absence in single desk offices.
They further tag activities based on this knowledge.
A new approach for occupancy recognition is going on by understanding
the relationships existing between carbon dioxide concentration and indoor air
quality IAQ in terms of occupant number. Physical CO2 model built on sensor
networks (Aglan, 2003) have been used extensively in smart office projects to
improve occupancy comfort and building energy use. However in this paper, CO2
physical model is studied to find out the valuable of using it in occupancy
estimation. However, for various applications like activity recognition, or context
analysis within a larger office space, information regarding the presence or
absence of people isn’t sufficient, and an estimation of the number of people
occupying the space is essential. (Lam et al., 2009) investigates this problem in
open offices, estimating occupancy and human activities using a multitude of
ambient information, and compare the performance of HMMs, SVMs and
Artificial Neural Networks. However, none of these methods generate human
understandable rules which may be very helpful to building managers.
In general, an occupancy count algorithm that fully exploits information
available from low cost, nonintrusive, environmental sensors and provides
meaningful information is an important yet little explored problem in office
buildings.
© ASCE
PROCESS USED FOR ESTIMATION
Experiment setup. The testbed is an office in Grenoble Institute of Technology,

which accommodates a professor and 3 PhD students. The office has frequent
visitors with a lot of meetings and presentations all through the week. The setup
for the sensor network includes:
• 2 video cameras for recording real occupancy numbers and activities.
• an ambience sensing network, which measures illuminance,
temperature, relative humidity (RH), motion at a sampling rate of 30 seconds.
• a centralized database with a web-application for
retrieving data from different sources continuously.
All the data possibly used for estimating the occupancy are called features
as in machine learning.
Generating and Selecting features. The underlying approach for the

experiments is to formulate the classification problem as a map from a feature
vector into some feature space that comprises several classes. Therefore, the
success of such an approach heavily depends on how good (those which provide
maximum separability among classes) the selected features are. In this case,
features are attributes from multiple sensors accumulated over an interval. The
choice of interval duration is highly context dependent, and has to be done
according to the granularity required. However, some features do not allow this
duration to be arbitrarily small. As an example, it has been observed that
levels do not rise immediately, and one of the factors affecting this time is the
ventilation of the space being observed. Regarding the results presented in this
paper, an interval of Ts = 30 minutes (which has been referred to here as 1
quantum) has been considered. Before any features are calculated for the training
data, some basic preprocessing of data had to be done: basic interpolation for
nonexistent data and application of an outlier removal algorithm. The
interpolation part is necessary for filling in missing values from the sensor data.
This is frequent in devices which are event-triggered i.e. no data points are
reported if there is no change in the feature being reported. Thus, the previous
data point had to be copied into the voids. One quantitative measurement of the
usefulness of a feature is information gain. Before detailing what is an
information gain, it is imperative to discuss the concept of entropy. Entropy is an
attribute of a random variable that categorizes its disorder. Higher the entropy,
higher is the disorder associated with the variable i.e. the less it can be predicted.
Mathematically, entropy is defined by: ( ) = ∑ − ( ) ( ) where y
is a random variable whose value domain is ( )= ,…, , H(y): is the
entropy of a random variable y and ( ) is the probability for y to be equal to the
value . Information gain can now be defined between two random variables, x
and y as: ( , ) = ( ) − ( | ) where y is a target random variable, H(y) is
the entropy of y and H (y|x) is the conditional entropy of y for given x. The higher
the reduction of disorder by fixing feature x is, the more is the information gained
for determining y thus making x a good feature to use for classifying y.
Learning process. From the large set of features displayed in, (Abhay Arora and
Bandyopadhyay, 2015), some of them may not be worthwhile to consider, to
achieve our target of occupancy classification. These features are ones, which
© ASCE
when added to the classification algorithm make no difference to overall output.

Regarding the mathematical calculation of the information gain, which was
discussed in, (Abhay Arora and Bandyopadhyay, 2015). Let Tk = ti: ti [kTs, (k +
1)Ts] be the time samples related to time quantum k, the most relevant features
are:
1. Fluctuation count: The PIR sensor in use is a binary sensor that
reports a value of 1 whenever it senses some motions. The number of
times a motion is detected within the specified duration of 1 quantum has
been computed. ∑ ( ).
2. Occupancy from power consumption, which estimated for 4 sensors
by =∑ with ∈ 0,1 . It satisfies: if poweri < threshold
then = 0 else = 1 where poweri stands for the actual laptop average
power consumption during time quantum i and threshold = 15W
3. Average: | | ∑ ( ) where 0 ≤ ≤ 47 for one day, since
the number of ‘half-hours’ in a day are limited to 48. This feature is
calculated for carbon dioxide.
4. Time slot generated from calendar. One of NIGHT, PRELUNCH,
LUNCH, POSTLUNCH. These correspond to time intervals [20-8), [8-
12), [12-14), [14-20) respectively.
5. First order derivative: Gives the trend of data. The data points are
interpolated to a first-order linear equation, and then the derivative of the
resultant line is recorded. This feature is useful to quantify the rate of
increase / decrease of occupancy relative to the previous time interval.
This feature is calculated for carbon dioxide.
6. Contact state: this feature is extracted for the door contact sensors.
Possible values for this feature can be 0: door open, 1: door closed, a real
number ∈ 0,1 , which denotes state change.
In this paper new features are added in addition to the previous one: Audio
microphone detection and occupancy from physical model. The correlation
with occupancy estimation with these features is discussed below.
Classification algorithm. A supervised learning approach has been used.

Occupancy has been measured before using a classification algorithm. Occupancy
count was manually annotated using the video feed from two cameras
strategically positioned in the office. The decision tree classification technique has
been selected because both it provides very good results and the results are easy to
read, analyze and adapt. The decision tree algorithm selects a class by descending
a tree of decision nodes. Each internal node represents a comparison of a single
feature value with a learnt threshold. The target of the decision tree algorithm is to
select features that are more useful for classification. One quantitative
measurement of the usefulness of a feature is the information gain that has been
discussed in (Abhay Arora and Bandyopadhyay, 2015). As information gain
approaches to zero, the difference between initial disorders (entropy) of the target
variable, and after having added knowledge from the test feature x is negligible.
Hence, the particular feature is not probably going to help very much during the
decision making process. Decision tree algorithm provides quite a few
advantages. As per (Quinlan, 1986), the features with higher information gain are
much higher up the tree, therefore making the process of feature selection intrinsic
© ASCE
to the classifier. Since the path to the leaf may consist of many internal nodes,
each of which may check different feature values, such paths exploit the
correlation among the various features. The decision tree approach offers the
advantage of generating rules that the path towards the leaf node is quite
informative and it clearly points out direct causes for the selection of a particular
class. Unlike methods that use decision boundaries (SVMs, regression
techniques), decision tree analyses are independent of the scale of the input data,
so no conditioning of the data is necessary.
Using this raw training data, previously mentioned features were
extracted. A vector of features and target 〈 , , , … , ; 〉had been generated
for each time quantum, where fi stands for the ith feature and y, for the level of
occupancy.
Occupancy from Acoustic Sensor. Acoustic features are a very important part of
occupancy classification when other non-intrusive sensors offer low class
separation. A single omnidirectional microphone can be used as an important tool,
when it comes to classify occupancy. Omnidirectional microphones are ones,
which can pick up sound from virtually any direction. They are considerably
cheaper than having multiple unidirectional microphones, and prove to be much
advantageous in places where it is required to track/ listen to multiple sources like
in meetings, discussions, (Abhay Arora and Bandyopadhyay, 2015). In this paper,
the recording signal from an office is generally background environmental noise
with a few human voices, some door opening, and tapping events. From the
recording signal RMS amplitude feature is defined, which is the root mean square
(or average) of the amplitude of a sound. However, it is related to the volume of
∑ ( )
the sound: = , where n is the number of samples taken and Si the
th
i sample. High and low RMS value will give indicator to the level of occupants
inside the office, this relationship is easy to visualize in (figure 2, left side), which
represents both, the RMS amplitude in dB for 4 days, and the actual occupancy
profile with respect to time (quantum time is 30 minutes).
Occupancy from physical model. An alternative approach for occupancy

estimation can be done by using physical model. According to ASHRAE
(1985), the model given by (1) represents the relation between carbon dioxide
generation, the volumetric flow rate of fresh air entering the office, the volumetric
air flow rate outgoing from the office and occupancy (Aglan, 2003). The proposed
approach relies on the data coming from concentration sensors, door contact,
window contact, occupancy labels extracted from video cameras for tuning air
flows, and constant parameters associated to the office.
( )
=− ( )+ ( ) ( )+ ( ) + ( ) ( )+
( ) (1)
It yields the following estimator:
, , ( ) ,
= − (2)
( )
= and =
Where:
• time quantum Ts =1800 seconds.
• indoor concentration: ( )
© ASCE
Table 1: Adjusted parameter values for physical model

Parameter Initial value Adjusted value
S 7 ppm. m3/s 19.6 ppm. m3/s
395 ppm 420 ppm
0.004 m3/s 0.076 m3/s
0.004 m3/s 0 m3/s
0.04 m3/s 0.1 m3/s
Table 2: Levels of occupancy considered with ranges

Number of Discretizations
levels
L=2 =0, >0
L=3 = 0 , > 0, ≤ 3 , > 3
L=4 = 0 , > 0, ≤ 2 , > 2, ≤ 4 , > 4
L=5 = 0 , > 0, ≤ 1 , > 1, ≤ 2.2 , > 2.2, ≤ 3.2 , >
3.2
L=6 = 0 , > 0, ≤ 1 , > 1, ≤ 2 , > 2, ≤ 3 , > 3, ≤
4, >4
• corridor concentration: ( )
• average opening of the door during a time quantum k: ∈
0,1
• production for 1 average person: S
• number of persons:
• air flow exchange with corridor: , = + 0 where
stands for leak air flow with corridor and window air flow is
assumed to be proportional to door opening
The first step is to find the best parameter values for invariant parameters
S, , , and QD using an iterative nonlinear optimization approach,
taking into account the positions of the door and the window, as shown in table 1.
An objective function is determined to minimize the difference between actual
and measured number of occupants in the room. Optimization covers a long
period of time but it can be imagined that less representative observations could
be sufficient.
The next step is to use these adjusted parameters for calculating the
number of occupants over a time quantum lasting 30 minutes. Occupancy
estimation is obtained from equation (2). Finally, the last step is to use this
estimation of occupants as one feature in the classification model.
Deciding the number of occupancy levels. In this section, how to choose the
number of levels (L) of occupancy for classification is discussed. This number is
not fixed and can be changed according to the required average error (average
distance between actual occupancy numbers and the mid points of estimated
levels). To determine the number of levels and related non overlapping ranges of
occupancy, training data are partitioned into L clusters with 2 ≤ ≤ , where
N is the maximum possible number of occupants. At L = 2, the problem amounts
© ASCE
too classify presence aand absencce of peopple. Table 2 shows the differeent
diiscretization
ns considereed (N = 4).
Table 3: Bassic Set OF Features

F
Basicc Set OF Feaatures
1. motionn detector co
ounting
2. occupaancy estim mation frrom pow
wer
consuumption
3. avverage valuee
4. time sllot
5. deerivative
6. door poosition
Fiigure 1: (lefft) Occupaancy estimaation consid

dering basicc features
(right) Occup pancy estim
mation considering alll the featurres
R
RESULTING
G OCCUP
PANCY EST
TIMATOR
RS
The C4.5
C decisioon tree algoorithm has been
b used to
o perform reecognition by
b
ussing aggreggated featurees and the labels extraacted from video cameeras. Traininng
daata cover 11 days from 4 May 2015 to 144 May 20155 while tessting data are a
coollected oveer for 4 dayys from 17 May 2015 to 20 May 2015. Oveer the traininng
peeriod, 12000 00 data poinnts have beeen collectedd.
Figurre 1, left sidde, shows thhe result obbtained fromm the learnt decision trree
coonsidering the
t basic sett of featuress (table 3), aas input to the
t detectio on model. The
T
pllot shows both
b actual occupancy
o profile and the estimaated profile as a graph of
nuumber of occupants
o w
with respectt to time (qquantum tim me is 30 minutes).
m T
The
avverage errorr yields to 0.32
0 occupannt.
Figurre 1, right siide, shows the result oobtained froom the decission tree aft
fter
coonsidering the two additional
a f
features of audio miccrophone detection
d annd
occcupancy fromCO2
fr phhysical moodel, in adddition to th he previous basic set of
feeatures. CO2 2 average value
v and CO2
C moved from the initial set
derivatiive are rem s
off features annd replaced by the estim mation of occupancy frromCO2 ph hysical moddel,
eqquation (2). Consideriing these features,
f leaads to impprovement in i occupanncy
esstimation with an averaage error of 0.24occupaant.
Both acoustic preessure (figu ure 2, left sidde) and occcupancy from
m CO2
phhysical mod del (figure 2,
2 right side)) are observved to be on ne of the moost importannt
feeatures for occupancy
o c
classificatio n, accordingg to the finaal Decision tree
n which rankks the featurres ascendinngly due to informationn gain for
cllassification
eaach feature, (figure 3,riight side). Acoustic
A preessure improoves the esttimation in
© ASCE
occcupancy att high levelss while occuupancy fromm CO2 physsical model decrease thhe
w
whole averag ge error in thhe classificaation.
Finallly, (figure 3, left side) show ws the ressults of average
a errror
coorrespondinng to each leevel. Accorddingly, 5 levels of occuupancy is thhe best option
foor the occuppancy classification.
Figure 2: (left) Correlation between acoustic prressure and

d occupatioon
(right) estimation ofo CO2 usin
ng a physiccal model
Fiigure 3: (leeft) resultin

ng estimatioon error function of number
n of
occupan
ncy levels (rright) Norm malized Infformation Gain
G from final DT
CONCLUSIIONS
A suppervised leearning appproach havee been prooposed in this t paper to

esstimate the number of occupants in i a room. InI the preseented appliccation, motion
fluuctuation counters using PIR sen nsors, poweer consumpttion sensorrs, CO2 meean
annd derivativve and dooor positionn is the m most intereesting inforrmation. TheT
esstimation off the numbeer of occup pants using a physical CO2 modeel is also veery
prromising. Classificatio
C n has beenn done usingg the C4.5 classificatioon algorithm m,
w
which leads to decisioon trees. Application
A to an office leads too an averaage
esstimation errror of 0.24 occupant foor 4 days peeriod, whichh is quite goood.
Superrvised learnning has beeen done thhanks to 2 video cam meras but thhis
appproach is limited becaause of privacy issues. Another op ption has been envisageed:
ussing discrette feedbackks from occcupant them mselves such h as with a keyboard or
anny other meeans. In adddition, becauuse decisionn trees are human
h readaable, they can
c
bee adjusted using
u expertt knowledgee, adjusting threshold for
f instance, or removinng
soome nodes when an innformation is not available, depen nding on th he considerred
livving areas. The two extensions
e can be com mbined to avoid the use of viddeo
caameras. It will
w be invesstigated furtther in the future.
fu
R
REFERENC CES
A
Abhay Aroraa, Manar Amayri,
A V.. R. B. S. P. and Baandyopadhyyay, S. 20115.
Estim
mating occuupancy in an office setting. 1BS-20155 Secretariat,
Hyderabad, Indiaa.
© ASCE
Agarwa l, Y., Balaji, B., Gupta, R., Lyles, J., Wei, M., and Weng, T. 2010.
Occupancy driven energy management for smart building automation. In
Proceedings of the 2nd ACM Workshop on Embedded Sensing Systems for
Energy-Efficiency in Building, pages 1–6. ACM.
Aglan, H.2003. Predictive model for co2 generation and decay in building
envelopes. JOURNAL OF APPLIED PHYSICS, 93(2).
ASHRAE, Atlanta, G. 1985. Fundamentals American Society of Heating,
Refrigerating and Air-Conditioning Engineers. Fundamentals American
Society of Heating, Refrigerating and Air-Conditioning Engineers.
Erickson, V. L., Carreira-Perpinãń, M. A´., and Cerpa, A. E. 2011. Observe:
Occupancy-based system for efficient reduction of hvac energy. In
Information Processing in Sensor Networks (IPSN), 2011 10th
International Conference on, pages 258–269. IEEE.
Kashif, A., Dugdale, J., and Ploix, S. 2013. Simulating occupants’ behaviour for
energy waste reduction in dwellings: A multi agent methodology.
Advances in Complex Systems, 16:37.
Lam, K. P., Höynck, M., Dong, B., Andrews, B., shang Chiou, Y., Benitez, D.,
and Choi, J. 2009. Occupancy detection through an extensive
environmental sensor network in an open-plan office building. In Proc. of
Building Simulation 09, an IBPSA Conference.
Milenkovic, M. and Amft, O. 2013a. An opportunistic activity-sensing approach
to save energy in office buildings. In Proceedings of the fourth
international conference on Future energy systems, pages 247–258. ACM.
Milenkovic, M. and Amft, O. 2013b. Recognizing energy-related activities using
sensors commonly installed in office buildings. Procedia Computer
Science, 19:669 677.
Nguyen, T. A. and Aiello, M. 2012. Beyond indoor presence monitoring with
simple sensors. In PECCS, pages 5–14.
Padmanabh, K., Malikarjuna V, A., Sen, S., Katru, S. P., Kumar, A., Vuppala, S.
K.,Paul, S., et al. 2009. isense: a wireless sensor network based conference
room management system. In Proceedings of the First ACM Workshop on
Embedded Sensing Systems for Energy-Efficiency in Buildings, pages 37–
42. ACM.
Page, J., Robinson, D., and Scartezzini, J. 2007. Stochastic simulation of occupant
presence and behaviour in buildings. Proc. Tenth Int. IBPSA Conf:
Building Simulation, pages 757–764.
Philipose, M., Fishkin, K. P., Perkowitz, M., Patterson, D. J., Fox, D., Kautz, H.,
and Hahnel, D. 2004. Inferring activities from interactions with objects.
Pervasive
Computing, IEEE, 3(4):50–57. Quinlan, J. R. 1986. Induction of decision trees.
Machine learning, 1(1):81–106.
Robinson, D. and Haldi, F. 2009. Interactions with window openings by office
occupants. Energy and Buildings, 44:2378–2395.
Roulet, C., Fritsch, R., Scartezzini, J., and Cretton, P. 1991. Stochastic model of
inhabitant behavior with regard to ventilation. Technical report.
© ASCE
Event-Based Parallel Simulation with a Sensing System for Occupant

Distribution Estimation in the Whole Building Scale
Zhenning Lang1 and Qing-Shan Jia1,*

1
The Center for Intelligent and Networked Systems, Department of Automation,
Tsinghua National Laboratory for Information Science and Technology, Tsinghua
University, Beijing 100084, China. E-mail: langzn13@mails.tsinghua.edu.cn;
jiaqs@tsinghua.edu.cn
*
Corresponding author.
Abstract
The information of occupant distribution in the whole building scale is

important for the evacuation in emergency and the energy conservation of HVAC
systems. This information is usually estimated by a sensor network system. However,
the evaluation of the performance of the sensor network system is usually hard to be
achieved beforehand. Furthermore, the optimization of the sensing system design is
much more difficult. For this reason, a simulation platform is of great practical interest
for the purpose of the sensing system design and optimization. To this end, we
proposed an event based simulation platform which different human movement
models, sensor models and estimation algorithms could be input into. For the purpose
of simulation efficiency, parallel simulation is applied in our platform. A serious of
experiments show the high efficiency and generality of our simulation platform.
INTRODUCTION
In modern buildings, the occupant distribution information is important for

energy conservation under normal conditions (Chao and Hu 2004) (Garg and Bansal
2000) and for evacuation in emergency (Chen et al. 2005). The problem of occupant
distribution estimation is to estimate the number of occupants in each zone of a
building. This problem is usually solved by a single occupant sensor (Song et al. 2008)
(Wang et al. 2012) or a sensor network (Wang et al. 2010). Sensor fusion methods are
also widely used in this field. The performance of a sensing system is crucial for
evacuation and energy conservation. However, neither the evaluation of different
sensing systems nor the evaluation of different estimation algorithms is easy to be
achieved beforehand due to the randomness of sensor errors and occupant movements
in nature. In practice, we need some methods to help the design of a sensing system.
Multiple times of evaluations or the gradient information of the evaluation needs to be
obtained in advance. For this reason, an efficient simulation platform for the sensing
system of occupant distribution estimation is in great need.
In our paper, we proposed an event based parallel simulation platform for the
sensing system of occupant distribution estimation in the whole building scale,
especially in high rise buildings. Our simulation platform mainly owns two
advantages,
© ASCE
a) High efficiency. The existing simulation platform is usually time driven.

However, as shown in (Cassandras and Lafortune 2008), it could bring more benefits
to simulate the man-made systems using event-driven method. In addition, we
implemented the parallel simulation technology to improve the computing speed
further more.
b) Generality. In our simulation program, the simulation procedure of a
sensing system for occupant distribution estimation is divided into four parts, which
are a) the human movement simulation, b) sensor detection simulation, c)
estimation and d) evaluation. We reviewed the related researches and found that
these parts were studied separately. Our simulation platform is a modularized
program which consists of a backbone framework and some plug-in components.
The backbone framework provides the main procedure and connections between the
aforementioned four parts. The plug-in components are the detailed models or
algorithms, and provide interfaces for the users to specify their own models and
algorithms.
As mentioned above, the models and algorithms in our simulation platform
could be specified by the users. So we reviewed the occupant movement models,
sensor detection models and estimation algorithms, which provide a thorough
comprehension of the problem. The human movement models are mostly studied in
the energy conservation area, and it is mainly divided into three categories, which are
deterministic model in (Abushakra et al. 2001), multi-agent model in (Gunathilak et
al. 2013) and (Liao and Barooah 2010) and Markov Chain model in (Wang et al.
2005), (Hutchins et al. 2007) and (Page 2008). The sensors used in the field of
occupant distribution estimation includes video cameras, infrared beam sensors, CO2
sensors and so on. The result of video cameras is usually modeled as a Gaussian
distribution in the field of image processing, such as in (Wang et al. 2010). In Wang
et al. (2012), the author assumes the result of an infrared beam sensor suffers from
the miss detection error and the false alarm error, and these two kinds of error are
both Bernoulli distribution. There are already some studies about the occupant
distribution estimation algorithms. In Wang et al. (2010) the author provided an
approximate estimation algorithm based on Bayesian estimation. In Meyn et al.
(2009) the author proposed a general framework for the occupant distribution
estimation sensing fusion problem. Machine learning and data mining methods are
also used by some researches on this problem, such as in Ekwevugbe et al. (2013).
The rest of the paper is arranged as follows. In part II the architecture of our
simulation platform is introduced in detail. Some experiments are used to show the
high efficiency and generality of our simulation platform in part III. Then the last
section is a conclusion.
INTRODUCTION TO THE SIMULATION PLATFORM
To introduce the simulation platform more clearly, we firstly formulate the

problem of occupant distribution estimation mathematically in part III.A. Then the
details of the simulation platform are introduced in part III.B.
© ASCE
A. Problem Formulation. As aforementioned, our simulation platform consists of

four parts, which are occupant movement simulation, sensor detection simulation,
estimation and evaluation. In our paper the formulation of the estimation and
evaluation parts is not described in detail for the reason that our work provides a
framework instead of detailed estimation and evaluation methods. A quick
introduction of these two parts is given in part III.B. In this part we focus on a general
formulation of the occupant movement and sensor detection.
Figure 1. Graphical Model of the Floor Plan
1) The Occupant Movement: Given the floor plan of a building, it is divided

into different zones, and one zone does not overlap with the others. A zone could be a
room, a corridor, or a part of them. If an occupant can walk from one zone to another,
it is defined that there is a boundary between these two zones. The left part of Fig. 1 is
a floor plan example, which includes several rooms and a corridor. The black labels
and red labels indicate different zones and boundaries respectively. A floor plan with
zones and boundaries could be naturally formulated as an undirected graph
= ( , ) as shown in the right part of Fig. 1, where = { , = 1, … , }
represents the vertices (zones) and = { , = 1, … , } represents the edges
(boundaries). The time is discretized as = {1 , … , , … , } , where is the
largest simulation time given by the user. The distribution of occupants at time instant
is denoted as a vector = { , … , } , where represents the number of
occupants in zone at time , and means transpose operation. Assume that the
maximum number of occupants in the simulation is , and each occupant owns an
unique ID. At time , is used to denote the zone which occupant is in. Then the
vector = ,…, represents at time where the occupants are. It should be
noticed that can be calculated from by the expression =∑ , where
is the indicator function. However, we cannot get from , which means that
contains more information than .
It is assumed that a movement between zones may happen between two
successive time instants. Then between time and + 1 , we assume that the
occupants move at time + 0.5. Correspondingly, define a boundary movement
matrix,
, , ,
. . … .
. = … … ⋱ …
, , ,
. . … .
© ASCE
,
where . denotes the movement from zone to zone between time and
+ 1. . is calculated based on the user specified human movement model and
the state vector . The update equation of is = + . , in which
is a n dimensional column vector with all one entries.
2) The Sensors: It is assumed that there are only two kinds of sensors, namely
in-zone sensors and boundary sensors. Although there are a plentiful types of
different sensors which are capable of providing the information about occupants,
most of these sensors may fall into these two types.
The in-zone sensors can provide the information of the number of occupants
in a zone. At time , the map from the distribution vector = { , … , } to the
observation value is formulated as a probability model. The observation value of
in-zone sensors is denoted as ={ ,…, } , where represents the
observation of the in-zone sensor in zone at time . In our simulation model, it is
assumed that the probability distribution of only depends on . To put the
formulation more clearly, the video camera is used as an example to show our model
of the in-zone sensors. In Wang et al. (2010), the estimation value of the number of
occupants in a zone given by a video camera is modeled as a sum of the true number
of occupants and a Gaussian noise. Generally speaking, the observation of an in-zone
sensor is a random variable, and its distribution depends on the number of occupants
in the zone. So is a function of , and the random vector , namely
= ( , ). The function is specified by the user, using the interface provided
by our platform.
The other type of sensors is called the boundary sensor. When an occupant
goes across the boundary, the boundary sensor provides related information. The
boundary beam sensor in Song et al. (2008) and the camera video in Chen et al.
(2008) are both widely used boundary sensors. From time to + 1, a boundary
sensor observation matrix
, , ,
. . … .
. = … … ⋱ …
, , ,
. . … .
is defined, which is corresponding to the boundary movement matrix . . The

, ,
observation of the boundary sensor is denoted as . . The distribution of .
,
is assumed to depend only on . just as the assumption for the in-zone sensors.
© ASCE
Figure 2. Flow Chart of the Simulation
B. The Architecture of the Simulation Platform. The architecture of the platform is

shown in Fig. 2. The yellow rectangles represent different simulation modules, and the
purple cylinders represent data. These two parts compose the Backbone Framework.
In our platform, the codes of the Backbone Framework cannot be specified by the user.
In Fig. 2 the green and dark green rectangles are the Plug-in Components which are the
interfaces to the users.
The Occupant Movement Simulation module in Fig. 2 is used to generate the
data of Occupant Movement Information and Real Occupant Distribution, namely the
boundary movement matrix . and the distribution vector for all the time
instants from the start time (1 ) to the end time ( ). Mathematically, the Occupant
Movement Information is a matrix set { . | = 1, … , − 1} and Real Occupant
Distribution is a vector set { | = 1, … , }. At the end of the Occupant Movement
Simulation module, { } and { } are stored on disk. In this module the Event Based
Simulation method is used, and this method is introduced in detail in (Cassandras and
Lafortune 2008). The algorithm for this module is shown in Alg. 1. is taken as the
system state, and its transition is triggered by the occupant movement event. As shown
in (Cassandras and Lafortune 2008), the essential part of an event driven simulation is
the maintenance of the ”SCHEDULED EVENT LIST”. In the “SCHEDULED
EVENT LIST” of our simulation, the event is defined as the occupant movement event.
So the length of the list is . The happening time of an occupant movement event is
given by the Occupant Movement Model.
Algorithm 1: Event Based Simulation of the Occupant Movement

Input: The graph model of the building , simulation time , , ,
Occupant Movement Model
Output:{ }, { } and { }
1: Initial the ”SCHEDULED EVENT LIST”, set = 0
2: Read and remove the entry ( , ) from the ”SCHEDULED EVENT
LIST”, where is the smallest of all the time in the list.
3: Set = , if > then exit, else continue.
4: Update and store . , and . (The update function has been
shown in the mathematic formulation)
5: Generate the new event ̃ and its time according to the Occupant
Movement Model, add ( ̃ , ) to the ”SCHEDULED EVENT LIST”.
© ASCE
Then { } and { } are inputt into the modules m Booundary Sensor Detecction
Sim
mulation andd In-zone Sensor
Se Detecction Simullation respectively as shown
s in Fiig 2.
Theen the Bou undary Sennsor Observvation Dataa { . | = 1, … , 1} and the
In-zzone Sensor Observattion Data { . | = 1, … , 1} are gen nerated by this
module.
Based on the asssumption inn part III.A A that the observation ns of diffeerent
sennsors are independent, the simulaation proceddure of the sensors coould be diviided
intoo smaller sim
mulation prroblems. In each sub-pproblem, onlly a part of the sensorss are
mulated. Thiis property of data sepparability makes
sim m it eaasy to simu ulate the sennsor
deteection in paarallel. In our
o platform m, ”Matlab Parallel
P Com mputing Tooolbox” is used
u
to implement
i the parallell simulationn of the Booundary Sennsor Detecttion Simulaation
andd In-zone Seensor Detection Simulaation module.
The and are then innput to thee module Estimation
E t generate the
to
Esttimated Occcupant Disttribution daata. In the eend, this daata and the Real R Occuppant
Disstribution daata are inpuut into the Evaluation
E module to generate thhe Performaance
dataa of the seensing system. In the Estimationn module, the estimattion algoritthm,
whiich is relateed to the occupant
o movement model
m and sensor
s detecction model, is
given by the users. In thhe Evaluatiion modulee, the valuees of severaal performaance
mettrics are prrovided, inccluding the mean erroor rate, fivee-number suummary off the
estiimation erroor, operationn cost of thee sensing syystem per yeear, etc.
EX
XPERIMEN
NTS
A. The Efficiency Imprrovement of the Simu ulation Plattform. A 1000 zones and
10000 occupantts scale probblem is testted to show w the improv
vement in efficiency
e ussing
parrallel simulaation. The result is shown in Fig. 3, and it sh
hows that wiith the increease
of tthe number of CPUs, thet computiing time deccreases inveersely propo ortionally. This
T
couuld be explaained by thhe Amdahl’s law. In Fig. F 3 we can c see thatt using paraallel
sim
mulation impproves the efficiency
e siignificantly.
F
Figure 3. Im
mprovemen
nt in Efficieency using Parallel
P Sim
mulation
© ASCE
B. A Case Stu udy. A case study on thhe 1st floor of FIT buillding, Tsingghua Univerrsity
shoows that ouur simulatioon platform m could be used to simulate diffferent kindss of
models and alggorithms. TheT 1st floor of the FIT T building contains
c 81 zones, andd the
movements off 200 occupaants from 8:00 8 am to 88:00 pm in a day are siimulated in our
case study.
Three kinds
k of sennsors are coonsidered inn the case study,
s whichh are the viideo
cammera sensor, the infraared beam sensor andd the passiv ve infraredd sensor. Inn the
sim
mulation, thee video cam mera is used to count thhe number of o occupantss in a zone, and
its rresult is moodeled as a sum of the true numbeer of occuppants and a Gaussian noise
N(0 0,0.5 ). Theen the probability of th he camera too get the truue number ofo the occuppant
is 668%. The infrared
i beaam sensor is used to monitor
m the boundary event.
e Whenn an
occcupant enterrs or leavess a zone, thhe infrared beam senso or could deetect this evvent.
How wever, it suuffers fromm a 20% probability
p o miss dettection erroor. The passive
of
infrrared sensoor detects thet in-zonee occupant movementt. It is asssumed that the
proobability of the sensor tot detect onne occupant is 0.75, an nd the probaability to deetect
morre that one of the n occcupants in th he zone is 1 0.25 .
Using the detectioon results of o these thrree kinds ofo sensors, three kindss of
estiimation algoorithms are used in this case studyy. The first algorithm, which is caalled
Naïïve Estimattion, uses thhe results ofo the infrareed beam sennsors only. When a sennsor
deteects an arrival or a depparture, thee estimationn number pllus one or minus
m one. The
second algoritthm is Infrared Trigggered Videeo Counting. It uses the t informaation
fromm both thee passive innfrared sen nsors and tthe video cameras.
c W
When a passive
infrrared sensorr detects thee in-zone movement
m inn a zone, theen the videoo camera inn the
zonne will be trigger
t to count
c the number
n of occupants
o in
i the zonee. This kindd of
cooordination policy
p is maainly for thee purpose off energy sav ving. The laast algorithm
m is
callled Naïve Estimation
E with Revission, and it uses all thee results of the three kiinds
of ssensors.
Figgure 4. Simu
ulation Ressult of the Case
C Studyy
© ASCE
The result of the case study is shown in Fig. 4, we could see that the more
number of sensors are used, the better the estimation result is, which accords with the
intuition. And this case study shows that our simulation platform could integrate
different models and different algorithms.
CONCLUSION
In this paper, we proposed an event based parallel simulation platform, which

is used to evaluate different sensing systems for occupant distribution estimation in
the whole building scale. The event based simulation and parallel simulation
technology are used to improve the efficiency of the platform. In addition, our
modularized platform provides flexibility for the users to specify their own human
movement model, sensor detection model and estimation algorithm. This simulation
platform may assist the engineer to select a better design, or to optimize the design of
a sensing system for occupant distribution estimation.
Our present simulation platform is coded by Matlab. In the future we would
use C++ and MPI to recode the platform to improve the efficiency further more. In
addition, we will review the existing works to select some default models and
algorithms, which would be integrated in our platform.
ACKNOWLEDGEMENT
This work is supported in part by the Tsinghua-UTC Research Institute for

Building Energy, Safety and Control Systems, and the United Technologies Research
Center, and is supported in part by the National Natural Science Foundation
(60704008, 60736027, 61021063, 61074034, 61174072 and 90924001).
REFERENCES
B. Abushakra, A. Sreshthaputra, J. Haberl and D. Claridge, “Compilation of diversity

factors and schedules for energy and cooling load calculations”, Tech. rep.,
Energy Systems Laboratory, Texas A and M University, 2001.
C. G. Cassandras and S. Lafortune, Introduction to Discrete Event Systems, 2nd ed.,
Germany, Springer, 2008, ch. 1, sec. 3, pp 26–43, 557–587.
C. A. H. Chao and J. S. Hu, “Development of a dual-mode demand control
ventilation strategy for indoor air quality control and energy saving”,
Building and Environment, vol. 39, no. 4, pp. 385–397, Apr. 2004.
C. H. Chen, Y. C. Chang, T. Y. Chen and D. J. Wang, “People Counting System for
Getting In/Out of a Bus Based on Video Processing”, presented at 8th
International Conf. on ISDA, Kaohsiung, Taiwan , Nov. 26–28, 2008.
X. Chen, B. Zhao and X. Li, “Numerical investigation on the influence of
contaminent source location, occupant distribution and air distribution on
emergency ventilation strategy”, Indoor and Built Environment, vol. 14, pp.
455–467, 2005.
T. Ekwevugbe, N. Brown, V. Pakka, and S. Fan, “Real-time Building Occupancy
Sensing Using Neural-Network Based Sensor Network”, presented at 7th
© ASCE
IEEE Conf. DEST, Menlo Park, CA, USA , Jul. 24–26, 2013.
V. Garg and N. K. Bansal, “Smart occupancy sensors to reduce energy consumption”,
Energy and Buildings, vol. 32, no. 1, pp. 81–87, 2000.
G. Gunathilak, A. Prasannakumar, N. Nazarian, H. Naeimi, “A Generalized Event
Driven Framework for Building Occupancy”, presented at Symp. on
Simulation for Architecture and Urban Design, San Diego, California, USA,
2013.
J. Hutchins, A. Ihler, P. Smyth, “Modeling count data from multiple sensors: a
building occupancy model”, presented at 2nd IEEE Int. Workshop on
CAMPSAP, St. Thomas, VI, Dec. 12–14, 2007.
C. Liao and P. Barooah, “An integrated approach to occupancy modeling and
estimation in commercial buildings, presented at ACC, Baltimore, MD, Jun.
30–Jul. 2, 2010.
S. Meyn, A. Surana , Y. Q. Lin, and S. M. Oggianu,“A Sensor-Utility-Network
Method for Estimation of Occupancy in Buildings”, presented at 48th IEEE
CDC., Shanghai, P.R. China, Dec. 16–18, 2009.
J. Page, D. Robinson, N. Morel and J. L. Scartezzini, “A generalised stochastic
model for the simulation of occupant presence”, Energy and Buildings, vol.
40, no. 2, pp. 83–98, 2008.
J. Song, Y. F. Dong, X. W. Yang, J. H. Gu and P. P. Fan, “Infrared Passenger Flow
Collection System Based on RBF Nueral Net”, presented at Proceedings of
the Seventh International Conference on Machine Learning and Cybernetics,
Kunming, China, Jul. 12–15, 2008.
D. Wang, C. Federspiel and F. Rubinstein, “Modeling occupancy in single person
offices”, Energy and Buildings, vol. 37, no. 2, pp. 121–126, Feb. 2005.
H. T. Wang, Q. S. Jia , C. Song, R. X. Yuan and X. H. Guan, “Estimation of
Occupancy Level in Indoor Environment Based on Heterogeneous
Information Fusion”, presented at 49th IEEE CDC., Atlanta, GA, USA, Dec.
15–17, 2010.
H. T. Wang, Y. L. Lei and X. H. Guan, “Estimation of Occupant Distribution by
Detecting the Entrance and Leaving Events of Zones in Building”, presented
at IEEE Conf. MFI, Hamburg, Germany, Sep. 13–15, 2012.
© ASCE
Effects of Variant Occupancy Transitions on the Energy Implications of

Setpoint/Setback Control Policies
Zheng Yang1; Ali Ghahramani1; and Burcin Becerik2

1
Innovation in Integrated Informatics Lab, Department of Civil and Environmental
Engineering, University of Southern California, 3620 S. Vermont Ave., KAP 217,
Los Angeles, CA 90089-2531. E-mail: zhengyan@usc.edu; aghahram@usc.edu
2
Innovation in Integrated Informatics Lab, Department of Civil and Environmental
Engineering, University of Southern California, 3620 S. Vermont Ave., KAP 224C,
Los Angeles, CA 90089-2531. E-mail: becerik@usc.edu
Abstract
In commercial buildings, more than 40% of the energy is consumed by

HVAC systems. There is usually a significant difference between energy demand and
energy consumption in buildings. It has been widely recognized that occupancy is
one of the most important factors impacting actual demands for HVAC systems.
However, it is still not clear how and when occupancy should be integrated with
HVAC systems at the building level for a long period of time. This is a complex
problem as occupancy is variant in nature, and each day occupants have different
presences. This paper systematically identifies the significance of variant transitions
of occupant presence on the absolute energy efficiency and the relative performance
of building level setpoint/setback control policies. An actual office building is used
for implementing the proposed methodology. The results showed that the variant
transitions of occupant presence have significant influences on the absolute energy
efficiency but not on the relative performance of different setpoint/setback control
policies.
INTRODUCTION
Research on energy management to reduce the overall building energy

consumption while maintain occupant thermal comfort is becoming increasingly
important. In 2013, buildings accounted for nearly 40% of the total energy use in the
U.S. (USDOE 2014). In commercial buildings more than 40% of the energy is
consumed by HVAC systems, which provide heating, cooling and ventilation to
individual zones to maintain comfortable and healthy indoor environments. It is
estimated that the majority of the HVAC systems in buildings are inefficient
(USDOE 2014; ACEEE 2013). There is usually a significant difference between
energy demand to maintain comfortable thermal conditions and energy consumption
to meet the loads. (Lütz 2012). It has been widely accepted that occupancy is one of
the most important factors to determine actual demands for HVAC systems, therefore,
matching HVAC control policies to actual occupancy is a feasible and reliable
© ASCE
approach to improve energy efficiency without sacrificing comfort and functionality

(Kwok et al. 2011).
Even though extensive research has been conducted to control HVAC systems
by only responding to actual occupancy, occupancy based HVAC control is still a
complex problem. There needs to be a systematic approach for understanding the
energy implications of occupancy on HVAC controls at the building level over a long
period of time, as occupancy is variant in nature and different for different days. This
paper provides a framework to systematically identify the significance of variant
transitions of occupant presence on the absolute energy efficiency and relative
performance of building level setpoint/setback control policies. An actual office
building is used for implementing the proposed framework. The paper is structured as
follows: First, we introduce the relationships between occupancy and HVAC system
control, and review the current occupancy based HVAC controls; Then, we describe
the objectives and methodology for analyzing the effects of variant presence
transitions on absolute energy efficiency and relative performance of setpoint/setback
control policies; Next, we present the implementation of the framework and the
findings. Finally, we discuss the limitations and conclude the paper.
OCCUPANCY AND HVAC SYSTEM CONTROL
Occupancy determines active conditioning requirements (heating/cooling

periods and thermal conditions). In general, the importance of occupants in a
building’s HVAC heating/cooling energy consumption can be broken down into two
categories (Figure 1): Occupancy in a building (how occupants occupy a building)
and occupant actions in a building (how occupants behave in a building) (Hoes et al.
2009; Yang and Becerik-Gerber 2014; Yu et al. 2011). Occupancy is defined as time-
sequenced occupancy changes for a specific space, including presence and number of
occupants. Occupancy results in heat gain as occupants continuously generate heat
due to their metabolisms and activities. Occupancy is also associated with the use of
building systems such as lighting, and appliances such as computers, which radiate
and add heat to the environment. Occupant actions also impact HVAC loads, for
example interactions with building elements, such as blinds, windows, and doors
(Tabak and de Vries 2010; Andersen et al. 2009). In commercial buildings, the
majority of the energy consumed by HVAC systems is used to meet the loads from
interior sources, such as heat gain and exterior sources, such as solar radiation. No
matter how significant the loads are, whether they are considered as actual demands
depends on the presence of occupants. If there is occupant in the zone, all the loads
are effective and should be met by the HVAC system. If a zone is unoccupied, some
of the loads are ineffective and it is not required for the HVAC system to meet the
full loads.
© ASCE
Figure 1. The relationships between occupancy and HVAC loads
Motivated by the significance of the relationship between occupancy and

HVAC loads, a range of research initiatives has been undertaken to optimize HVAC
controls based on presence of occupants (Erickson et al. 2011; Dong and Andrews
2009; Yang et al. 2012). The basic principle is that energy efficiency could be
improved by not fully running HVAC systems in vacant zones. Substantial energy
savings have been reported by prior research by not maintaining static setpoints in
unoccupied zones. Instead, zone temperatures were allowed to float within a certain
range (Agarwal et al. 2010; Oldewurtel et al. 2013; Yang et al. 2013). Despite the
high volume of research and the increasing number of commercial solutions in
occupancy based HVAC controls, it is still not clear how and when occupancy should
be integrated with HVAC systems. Some of the previous research generated case-
specific control solutions and may not be scalable to other types of buildings. This is
a complex problem as the transitions between occupied/unoccupied statuses are
variant and different for different days. There is no systematic understanding towards
the patterns of absolute energy efficiency and relative performance of
setpoint/setback controls under variant occupancy. A general framework is necessary
to systematically identify the significance of variant transitions of occupant’s
presence on energy implications of building level setpoint/setback control policies.
OBJECTIVES AND METHODOLOGY
Operation of an HVAC system is triggered as a response to the temperature

changes based on the setpoints. Setpoint is the terminal level temperature setting in
each zone. Setpoint regulates the desired temperature range (i.e., the deadband) and is
the primary parameter for terminal control. Since HVAC terminals respond to the
loads through the control of setpoints, the setpoint is commonly used as the medium
to implement occupancy based HVAC control for energy efficiency. During an
unoccupied period, allowing a setpoint to float to a different temperature, which is
defined as the setback, could potentially reduce heating/cooling loads on the demand
side. Occupied/unoccupied transitions are commonly considered to be equal to the
effective/ineffective load transitions, which means the loads are ineffective and not
© ASCE
necessary to be met when the zone is unoccupied. However, when a zone becomes
unoccupied, instant control adjustments may cause discomfort, as an occupant may
reoccupy the zone at any time and reconditioning from a deep setback to a setpoint
after a relatively short period of vacancy may consume more energy than just
maintaining the setpoint. Therefore, a terminal control could be divided into four
periods: (1) the setpoint period: a terminal works to maintain the setpoint, loads are
considered effective; (2) the float period: a terminal is off and the temperature floats
from setpoint to setback, loads are considered ineffective; (3) the setback period: a
terminal works to maintain the setback, loads are considered effective; (4) the
reconditioning period: a terminal works to restore temperature from setback to
setpoint, loads are considered effective (Figure 2). In sum, the transitions between
occupied/unoccupied statuses do not necessarily follow the transitions between
effective and ineffective loads. A portion of the loads during unoccupied periods
should be considered effective for improving energy efficiency.
As shown in Figure 2, the time lag and limit for temperature to float
determine the transitions of effective loads and ineffective loads. They are called
setpoint/setback schedules (e.g., waiting time to trigger setback) and distances
(difference between setpoint and setback) in this paper. Conceptually, different
combinations of setpoint/setback schedules and distances may lead to significantly
different levels of energy efficiencies. The synergetic effects of setpoint/setback
schedules and distances are investigated for energy efficiency. Energy efficiency in
this paper incorporates both energy reduction, which is the absolute amount of energy
savings, and the conditioning miss, which is the length of time during which a space
is occupied but the temperature is outside the comfort range. Conditioning miss is
considered as an equally important component as the energy reduction because it
compromises the basic function of an HVAC system to maintain a comfortable
thermal environment. 23oC is used as the static setpoint when the zone is occupied.
The interval for setpoint/setback distance is 2 K and the interval for setpoint/setback
schedule is 5 minutes. Both setpoint and setback have a deadband of ±1 K and the
conditioning miss is the length of time a zone is occupied but the temperature is
outside the range of [22oC, 24oC] in order to be in compliance with the PMV
(Ghahramani et al. 2015). It is important to note that, the ideal heating/cooling loads
are studied in this paper, thus the loads from HVAC components (primary/secondary
systems and terminals) are neglected. For example, an assumption used in this paper
is that there is no energy loss in the transportation of air. Ventilation is considered to
provide airflow per the ASHRAE requirement (ASHRAE 2010).
Figure 2. Deviation between occupancy and effective heating/cooling loads
© ASCE
Variant Occup pancy Tran nsitions and d Energy Effficiency. T The energy im mplications
of discrete com mbinations ofo setpoint/ssetback scheedules and ddistances aree simulated
usin ng whole buuilding energ gy simulation. The weigghted sums oof energy redduction and
connditioning miss
m represeent the energy efficienncy of diffe ferent combinations of
setppoint/setbackk schedules and distancees. Since thee presence oof occupants depicts the
onee-time occurrrences of occcupant preseence/absencee changes, thhe variant trransitions of
occcupant preseence mean the differences of occcupied/unocccupied trannsitions for
diffferent days. In order to analyze the effects of vaariant occuppancy transittions on the
eneergy efficienncy of setpoint/setback controls,
c twoo sets of sim
mulations are conducted:
1) is using actual occup pancy with variant traansitions (eaach day haas different
occcupancy), wh hile 2) is usiing repeatedd occupancy with constannt transitionns (each day
hass the same occupancy).
o There are N possible sscenarios foor the second set if the
perriod has N daays, and each h scenario iss called one sample repreesenting onee possibility
of occupancy
o happening
h duuring that peeriod withouut variant traansitions. Thherefore, the
firsst objective of
o this paperr is to test whether
w 1) annd 2) have ssignificant ddifference in
eneergy efficienncy of differeent setpoint//setback scheedules and ddistances. A process for
commparison is designed
d and shown in Figure3. Zoone level preesence is dettermined by
agggregating thee occupancy of rooms in n that zone: ((1) if at leasst one room in a zone is
occcupied, the zone is occup pied; (2) if alll rooms are vacant, the zone is unocccupied.
First thee energy effficiency resuults of differrent setpointt/setback conntrols using
actual occupan ncy are obtaained, and th he estimationn interval fofor each com mbination is
calcculated withhin ±2.5% off the simulatted energy effficiency givven the preddefined 95%
connfidence leveel. Then thee repeated occupancy is used to repplace actual occupancy,
andd the simulaation is ran for
f N times, each time uusing one’s occupancy for N days.
Forr each comb bination of setpoint/setb
s ack schedulle and distannce, the percentages of
eneergy efficien ncy levels of all sam mples (all ppossible sceenarios usinng repeated
occcupancy with hout variant transitions) within the iintervals from m the actuall occupancy
are calculated (it is calleed coverage percentagee in this paaper). The ssmaller the
covverage perceentage is, the more sign nificant the iinfluence thhe variant traansitions of
occcupant preseence on thee absolute energy efficciency of ddifferent schhedule and
disttance combinations.
Fig
gure 3. Comp
parison of en
nergy impliccations betweeen actual annd repeated occupancy
Variant Occup pancy Tran nsitions andd Relative PPerformancee. The seconnd objective
is to
t test whethher actual occcupancy with variant ttransitions aand repeatedd occupancy
witthout varian
nt transitionss have sign
nificant diffeerence in reelative perfoormance of
diffferent setp
point/setbackk schedules and diistances coombinations. Relative
perrformance is defined as the performance of one combinatioon for energyy efficiency
© ASCE
compared to the best and worst combination in the same scenario. The difference
between absolute energy efficiency and relative performance is: absolute energy
efficiency shows the actual energy efficiency that can be improved, while the relative
performance evaluates differences of selected combinations from the remaining. Even
if the absolute energy efficiency levels from repeated occupancy and actual
occupancy are different, the combinations of setpoint/setback schedule and distance
that outperform other combinations could still be the same (e.g. [20%,40%,60%,80%]
and [17%,29%,41%,54%] are significantly different, but they are similar after
normalization). Therefore, the simulated energy implications of discrete
setpoint/setback combinations are normalized by ∗ = to indicate the
relative energy performance of different setpoint/setback controls. The same process
of comparison analysis (Figure 3) is then applied to test whether there is significant
difference between actual and repeated occupancy in terms of the normalized energy
implications. Similarly, the coverage percentage is calculated, and the smaller the
percentage is, the more significant the influence the variant transitions of occupant
presence on the relative performance of different schedule and distance combinations.
IMPLEMENTATION AND FINDINGS
The test bed building used in this paper is the Ralph & Goldy Lewis Hall
(RGL), a typical educational building on the University of Southern California
campus near downtown Los Angeles, California. It is a three-story building with a
footprint of 3,735 m2, and contains 89 mechanically ventilated rooms. The 16 zones
on the third floor consist of 28 rooms and were monitored by 28 wireless sensor units,
each of which includes a number of ambient sensors such as temperature sensor and
CO2 sensor for modeling real-time occupancy (Yang et al. 2013). The ideal loads
(system energy loss is neglected) for the third floor during May 2013 to April 2014
were simulated to implement the proposed methodology. One year was divided into
four periods based on the outside temperature statistics (mean and standard deviation).
23oC (with 1 K as deadband) was used as the static setpoint to maintain a comfortable
thermal environment when the zone was occupied. When there was no occupant, six
different setback values (2 K as the interval) and six setback waiting time (with 5 min
as the interval) were combinated. In this paper, energy reduction and conditioning
miss were considered equally important, thus the energy efficiency was expressed as
50%*(energy reduction)+50%*(conditioning miss). The period of N days for testing
the energy significance of variant occupancy transitions was three-month long and
therefore there were four periods to analyze the consistency of findings.
For each period, the influences of variant occupancy transitions on the energy
efficiency of different setpoint/setback schedules and distances were first investigated.
Actual occupancy and repeated occupancy were used and statistics of coverage
percentages for all 36 combinations were shown in Table 1. The differences of energy
efficiency caused by variant occupancy transitions were around 60%, indicating that
the variant transitions of occupant presence had significant influence on the absolute
energy efficiency of different setpoint/setback schedules and distances. The
normalized energy implications of different setpoint/setback controls from actual
occupancy and repeated occupancy were then compared to investigate the influences
© ASCE
of variant occu upancy transitions on th he relative performancee of differennt setpoint/

setbback scheduules and distaances. From the statisticss of coveragge percentagees for all 36
commbinations in i Table 1,, it can be seen the ddifferences of normalizzed energy
effiiciency caussed by variant occupanccy transitionns were below 15%, inddicating that
the variant tran
nsitions of occcupant pressence did noot have signiificant influeence on the
relaative perforrmance of different
d seetpoint/setbacck schedulees and disttances. The
ran
nkings of com mbinations inn terms of ennergy efficieency remaineed the same..
Tab
ble 1. Coverrage percen ntages of eneergy efficien
ncy for all ccombination
ns
Eneergy Efficien
ncy Relative P Performancee
Mean
M SD Max. Min. M Meaan SD Maax. Min.
Maay-Jul (1) 45% 19% 75% 21 1% May-Juul (1) 94% % 3% 88% % 98%
Aug-Oct (2) 38% 15% 69% 20 0% Aug-Oct (2) 90% % 4% 85% % 95%
Nov-Jan (3) 42% 18% 71% 17 7% Nov-Jaan (3) 91% % 3% 87% % 97%
Feb
b-Apr (4) 40% 16% 73% 19 9% Feb-Appr (4) 92% % 3% 88% % 96%
The covverage perceentages of reelative perforrmances for four periodds were then
stattistically com
mpared usin ng paired T-tests. The results (Tabble 2) show wed that the
covverage percentages of relative peerformances were signnificantly diifferent for
diffferent periodds. Besides, the
t four periiods had diff fferent outsidde temperatuure statistics,,
andd it was foun nd that the difference
d off average temmperature beetween the ttwo periods
wass approximaately linearly associated d with the ddifference oof coverage percentage
betwween the saame periods.. The joint effects
e of ouutside temperature and the variant
occcupancy tran nsitions on energy
e efficciency were more signiificant than on relative
perrformance off setpoint/settback schedu ule and distaance combinations.
Tab
ble 2. Statisstical analyssis of covera
age percentaage differen
nces between
n periods
Relative Performance
P e
Compariison Sig.
Period 1-2
1 0.05
Period 1-3
1 0.02
Period 1-4
1 0.01
Period 2-3
2 0.00
Period 2-4
2 0.00
Period 3-4
3 0.03
CO
ONCLUSIONS
This paaper presenteed a framewwork to systeematically iddentify the ssignificance

of variant trannsitions of occupant
o preesence on thhe absolute energy effiiciency and
relaative perform
mance of buuilding level setpoint/setbback controll policies. A real world
offi
fice building g was used for implem menting the proposed fr framework. The results
sho
owed that thee variant traansitions of occupant
o preesence had ssignificant innfluence on
the absolute en nergy efficieency but did
d not have siignificant innfluence on the relative
perrformance of o different setpoint/settback scheddules and ddistances. T The outside
tem
mperature diffference was linearly asssociated wiith the effeccts of variannt occupant
© ASCE
transitions on both absolute energy efficiency and relative selection of different

setpoint/setback controls. However, there are limitations that will be addressed in our
future explorations. First only the actual occupancy in the test bed building was
considered and more diverse occupancy should be included to consolidate the
findings. In addition, the quantitative relationships between variant occupancy
transitions and energy implications of different setpoint/setback combinations have
not been investigated. Finally, long-term occupancy patterns could be considered to
categorize the variant transitions of occupant presence, which may follow certain
patterns.
ACKNOWLEDGMENTS

Foundation under Grant No. 1351701. Any opinions, findings, and conclusions or
recommendations expressed in this material are those of the authors and do not
necessarily reflect the views of the National Science Foundation.
REFERENCES
Agarwal, Yuvraj, Bharathan Balaji, Rajesh Gupta, et al. 2010. Occupancy-driven

energy management for smart building automation. BuildSys on Embedded
Sensing Systems for Energy-Efficiency in Building, Zurich, Switzerland.
ACEEE (American Council for an Energy-Efficient Economy). Commercial sector:
Buildings and equipment. 2013. http://www.aceee.org/portal/commercial.
Andersen, Rune Vinther, Jørn Toftum, Klaus Kaae Andersen, and Bjarne W. Olesen.
2009. Survey of occupant behaviour and control of indoor environment in
danish dwellings. Energy and Buildings 41 (1): 11-6.
ASHRAE (American Society of Heating, Refrigerating, and Air-Conditioning
Engineers). 2010. ANSI/ASHRAE standard 62.1-2010: Ventilation for
acceptable indoor air quality. ASHRAE, Atlanta, GA.
Dong, Bing, and Burton Andrews. 2009. Sensor-based occupancy behavioral pattern
recognition for energy and comfort management in intelligent buildings. The
11th IBPSA Conference, Glasgow, Scotland.
Erickson, Varick L., Yiqing Lin, Ankur Kamthe, et al. 2009. Energy efficient
building environment control strategies using real-time occupancy
measurements. First ACM BuildSys, Berkeley, CA.
Ghahramani, Ali, Chao Tang, and Burcin Becerik-Gerber. An online learning
approach for quantifying personalized thermal comfort via adaptive stochastic
modeling. Building and Environment 92 (2015): 86-96.
Hannes Lütz. 2012. The new version of EN15232 effects of building automation on
building efficiency. Honeywell Building Experts.
Hoes, P., JLM Hensen, MGLC Loomans, B. De Vries, and D. Bourgeois. 2009. User
behavior in whole building simulation. Energy and Buildings 41 (3): 295-302.
Kwok, Simon SK, Richard KK Yuen, and Eric WM Lee. 2011. An intelligent
approach to assessing the effect of building occupancy on building cooling
load prediction. Building and Environment 46 (8): 1681-1690.
© ASCE
Oldewurtel, Frauke, David Sturzenegger, Manfred Morari. 2013. Importance of

occupancy information for building climate control. Applied Energy 101:521-
32.
Tabak, Vincent, Bauke de Vries. 2010. Methods for the prediction of intermediate
activities by office occupants. Building and Environment 45 (6): 1366-72.
US Department of Energy. 2014. Building energy data book.
US Department of Energy. 2014. Commercial Building Energy Consumption Survey.
Yang, Zheng, Nan Li, Burcin Becerik-Gerber, and Michael Orosz. 2013. A
systematic approach to occupancy modeling in ambient sensor–rich buildings.
Simulation. 90.8: 960-977.
Yang, Zheng, Nan Li, Burcin Becerik-Gerber, and Michael Orosz. 2012. A non-
intrusive occupancy monitoring system for demand driven HVAC operations.
Construction Research Congress. West Lafayette, IN.
Yang, Zheng, Burcin Becerik-Gerber. 2014. The coupled effects of personalized
occupancy profile based HVAC schedules and room reassignment on building
energy use. Energy and Buildings. 78 (2014): 113-122.
Yu, Zhun, Benjamin Fung, Fariborz Haghighat, Hiroshi Yoshino, and Edward
Morofsky. 2011. A systematic procedure to study the influence of occupant
behavior on building energy consumption. Energy and Buildings 43 (6):
1409-17.
© ASCE
A Study of Time-Dependent Variations in Personal Thermal Comfort via a

Dynamic Bayesian Network
Ali Ghahramani1; Chao Tang2; Zheng Yang3; and Burcin Becerik-Gerber4

1
Innovation in Integrated Informatics Lab, Sonny Astani Dept. of Civil and
Environmental Engineering, Univ. of Southern California, KAP 217, 3620 South
Vermont Ave., Los Angeles, CA 90089-2531. E-mail: aghahram@usc.edu
2
Vermont Ave., Los Angeles, CA 90089-2531. E-mail: chaotang@usc.edu
3
Vermont Ave., Los Angeles, CA 90089-2531. E-mail: zhengyan@usc.edu
4
Environmental Engineering, Univ. of Southern California, KAP 224C, 3620 South
Vermont Ave., Los Angeles, CA 90089-2531 (corresponding author). E-mail:
becerik@usc.edu
Abstract
The current practice of defining operational settings for HVAC systems is to

use fixed set points, which assumes static and same comfort requirements for
building occupants. The majority of the research efforts in the literature study thermal
comfort through time-invariant learning algorithms. However, thermal comfort has
been shown to vary from person to person, and change over time due to climatic
variations or acclimation. In this paper, we present thermal comfort variation results
by studying the data from 33 human subjects and statistically evaluate and study the
variations to learn similarities and differences among these individuals. In order to
quantify the variations, we briefly describe our adaptive stochastic modeling
technique. The technique uses a systematic approach for detecting time dependent
thermal comfort variations for an individual. The results confirm that personal
comfort vary over time (average: 0.061 °C per day). In addition, we observed a high
standard deviation (0.159 °C) across the subjects’ preference variations.
INTRODUCTION
Commercial buildings account for about 18.9% of the total energy

consumption, and 19.59% of the total greenhouse gas emissions in the United States
(Book 2010; U.S. Energy Information Administration 2011). Heating, Ventilation,
and Air Conditioning (HVAC) systems have the largest impact on the energy usage
and gas emissions in commercial buildings (43% of the total commercial building
energy consumption (Book 2010; U.S. Energy Information Administration 2011)).
© ASCE
HVAC systems are responsible primarily for providing satisfactory thermal comfort
conditions. Thermal comfort is one of the most influential factors on overall
satisfaction with indoor environments (Frontczak and Wargocki 2011). Standard
models for thermal comfort conditions (e.g., ASHRAE Standard 55 (Thermal
Environmental Conditions for Human Occupancy) (Standard 2004)) use few selected
environmental and occupant related parameters (e.g., indoor air temperature, airflow
rate, activity, etc.) to predict occupants’ thermal comfort. In addition, standard models
are not adaptive to time dependent variations while individuals’ thermal comfort
ranges may change over time due to many dynamic environmental and human related
variables (Jendritzky and de Dear 2009). Moreover, it has been shown that humans
perceive comfort in a range of environmental thermal conditions (Nicol and
Humphreys 2002). The fact that adjusting the HVAC temperature set point by 1°C
have considerable impacts on the overall energy consumption (Hoyt et al. 2014; Yang
and Becerik-Gerber 2015) suggests that learning this range can potentially lead to
advanced control of building HVAC system for reducing energy consumption.
Human thermal comfort could be obtained using two types of approaches: (1)
survey based approaches; and (2) physiological measurement based approaches.
Survey based approaches use a participatory learning process through a
questionnaire, while physiological measurements based approaches utilize certain
physiological measurements (e.g., heart rate, skin temperature, etc.). Application of
these two categories of approaches requires real-time and continuous monitoring of
building occupants, making it a challenging task. In order to address this challenge, a
correlation analysis could be performed between instant comfort levels driven from
previously mentioned approaches with some other variables, such as environment
related variables (e.g., indoor air temperatures, clothing levels). Consequently, the
selected correlated variables could be used to estimate occupants’ thermal comfort
levels.
In this paper, we first describe our adaptive stochastic modeling approach for
modeling personalized thermal comfort of building occupants (Ghahramani et al.
2015). We then demonstrate the results of time dependent variations of thermal
comfort studying the data from 33 test subjects and statistically analyzing the results.
First, we provide a review of recent studies on personal thermal comfort learning and
modeling. Next, we describe our stochastic approach for modeling personalized
thermal comfort, as well as our time dependent variation analysis procedure. We then
explain the experimental design and present the results for time dependent variations
using the data from the subjects. Finally, we summarize the findings and conclude the
paper.
PERSONAL THERMAL COMFORT LEARNING AND MODELING
Thermal comfort is defined as the state of mind which expresses satisfaction

with the thermal environment (Standard 2004). Consequently, survey based
approaches potentially learn personal comfort more accurately than physiological
approaches as they try to directly extract the state of mind of a person. The major
challenge with both types of comfort learning approaches is that they require
continuous monitoring of or feedback from building occupants. To address this
challenge, statistical models have been utilized using correlation analysis between
© ASCE
few parameters that are easier to monitor (e.g., ambient factors) and occupants’
comfort. For example, in (Liu et al. 2007), a deep artificial neural network (ANN)
learning technique was used for classifying environmental conditions into
comfortable, uncomfortably warm, and uncomfortably cool. The ANN algorithm had
4 input variables (i.e., air temperature, radiant temperature, air flow, air humidity).
The algorithm was trained with comfort votes from subjects under controlled
experiments. However, similar to other studies in literature, the time dependent
variations of thermal comfort were not considered in this study. Researchers also
focused on adaptive thermal comfort modeling. For example, an adaptive technique,
which uses the PMV model as a prior model, was introduced in (Yao et al. 2009).
The model calculated an adaptation coefficient, which decreases or increases the
estimated PMV values. The adaptation coefficient took into account local climate,
culture, and social backgrounds.
Yet, the majority of the efforts in the literature lack the components for
detecting time dependent variations (preference variations over time due to hidden
variables) in thermal comfort. In a previous effort (Ghahramani et al. 2015), we
developed a modeling technique that enabled us detecting time dependent variations
in a systematic manner with no prior assumptions about occupant preferences. This
modeling technique used the data from occupants and the environment without any
pre-defined bias on the personal comfort. The modeling technique has an online
learning structure: when any time a new data point is communicated by an occupant
(i.e., comfort vote), we record the associated environmental condition and the model
gets updated. This learning process improved the accuracy of learning personal
comfort. The opportunity that this modeling technique provides is to mathematically
demonstrate the compliance of comfort predictions with the requirements stated in
standards (ASHRAE Standard 55). This work motivated us to study how personal
comfort varies over time and/or how the range of comfortable temperatures for each
person varies over time.
ADAPTIVE STOCHASTIC MODELING APPROACH
In this section, we briefly describe our comfort modeling approach that uses
internal parameters to capture the variations in an individual’s thermal comfort
preferences and explain how we quantify the variations, which is the focus of this
paper. The input data to the modeling approach consist of thermal comfort votes and
associated air temperatures. Thermal votes are divided into three categorical
variables: uncomfortably warm; comfortable; and uncomfortably cool. Figure 1
shows a sample dataset for an individual.
We first transform comfort votes/temperature data into a parametric
mathematical model. Upper Limit (UL) and Lower Limit (LL) for temperatures that
comfort can potentially be perceived by a subject and three probability distributions
for uncomfortable and comfortable conditions (LD – Lower Distribution, MD –
Middle Distribution, UD – Upper Distribution) are the parametric models that are
generated. UL is defined as the highest temperature that a subject has communicated
a comfort vote. LL is defined as the lowest temperature that a subject communicates a
comfort vote. LD, MD, and UD are the probability distribution functions defined for
the uncomfortably cool, comfortable, and uncomfortably warm data points,
© ASCE
resppectively. We
W then inteegrate the efffects of diffferent therm mal comfortt conditions
into
o a single jooint distributtion (JD) by y defining a Bayesian nnetwork to ccombine the
pro
obability disttributions ovver the rang ge of temperratures that evidence suuggests that
commfort can potentially
p be
b perceiveed [LL to UL]. A B Bayesian nettwork is a
pro
obabilistic graphical mo odel (a direected acyclicc graph) thhat representts a set of
ran
ndom variab bles and theeir condition nal dependeencies. The network inn our work
represents thee probabilisstic relation nships betw ween the influential probability
disttributions (L
LD, MD, an nd UD) and d the overalll comfort. A ASHRAE S Standard 55
(Staandard 2004 4) uses the PMV-PPD model to ddefine the reequirements for indoor
therrmal condittions. This standard requires that percentage of dissatisffied people
(PP
PD) to be leess than 20% %, which im mplies that aat least 80% % of the occupants in a
buiilding to be satisfied. Based
B on thee triangle innequality, if we set the probability
threeshold of ouur proposed approach to o 80%, the e xpected percentage of tthe satisfied
occcupants wo ould be grreater than 80%. Theerefore, thee ASHRAE E standard
req
quirements would
w be meet. Thus wee set probabbility threshoold (PT) as the rule of
classsification to
t be 0.8. This
T fact co
ould be usedd to alter thhe conventional multi-
objective (e.g.,, comfort an nd energy) optimizationn problems for buildinngs’ HVAC
systems by transforming comfort ob bjectives too constrain functions oof a single
opttimization problem
p as demonstrateed in ((Ghaahramani ett al. 2014; Yang and
Beccerik-Gerberr 2014)). Fig gure 1 illustrrates the diffferent compoonents of thee model.
Fiigure 1. Seg
gmentation of
o data and PT (Probab bility Thresshold) as a cconstraint
for classifying comfo
ort vs. discoomfort condditions
In ordeer to detect the time deependent coomfort (prefference) varriations, we

create a window w of data staarting at the most recentt data point and go backkwards (i.e.,
a sliding windo ow moving backwards)). We then iimplement tthe probabillistic model
exp
plained abov ve, and use a statistical teest (i.e., Kollmogorov–Sm mirnov test)) to detect if
the joint prob bability distrribution siggnificantly ddiffers from m the joint probability
disttribution thaat is generated from all data pointts. Kolmogoorov–Smirnoov test is a
nonnparametric test of the eq quality of coontinuous prrobability disstributions. IIt quantifies
a distance betwween the emp pirical distribbution functtions of two samples (Jaames 2006).
Thee details of the modelingg approach area provided in (Ghahram mani et al. 20015).
In this paper, we derived
d the range of teemperatures at which a subject is
mfortable following the standards (probability
com ( of being coomfortable ggreater than
80%%) through thet implemeentation of th he dynamic B Bayesian leaarning algorrithm on the
© ASCE
data for each subject. In the results section, we demonstrate the range for each
individual at every data point interval. In addition, we calculated the point that had
the highest probability of comfort and averaged over the duration of the experiment
for each subject, as well as the number of data points communicated. We also
calculated the variations in thermal preferences as absolute differences between the
previous and current most comfortable temperature values over an interval. The
intervals studied in this paper are data points or time (i.e., day). Consequently, we
calculated the data point-based variation analysis through calculating the difference in
red crosses in Figure 2 and averaged over all data points. For calculating the daily-
based variations, we multiplied the data point based calculation by the average data
points per day for each test subject.
EXPERIMENT PROCEDURE
The data collection was completed in several offices in the University of

California campus buildings. Based on the Köppen climate classification (Peel et al.
2007), the climate of the area is defined as a dry-summer subtropical climate (also
referred to as the Mediterranean climate). For such climates, the average temperature
in the warm months is above 10 °C and in the cold months is between -3 and 18 °C
(Peel et al. 2007). The test subjects were students, staff, and the faculty in the USC
campus buildings. Each test subject was given an ID number and asked to
communicate his/her votes. The temperature/humidity sensor, used in the
experiments, was Aosong AM2302 temperature/humidity sensor, which has an
accuracy of ±0.5°C for temperature and ±2% RH (Relative Humidity) for humidity,
and the resolution of 0.1°C for temperature and 1% RH for humidity. The sensors
where placed in a closed proximity (less than 1.5 meters to 2 meters) to the subjects.
The subjects were asked to communicate their votes while continuing their regular
office activities with at least 15 minutes time intervals. The number of data points and
the duration of data collection are presented in Table 1.
RESULTS
Following our methodology, we calculated the comfortable temperatures for

six-sample subjects over the duration of the experiment (Table 1) on a data point
interval and presented in Figure 2. Blue lines show comfortable temperatures and red
crosses show the temperature with the highest probability of comfort. As it can be
seen, there are points that there are no comfortable temperature ranges for an
individual. However, the algorithm can still detect the temperature point has the
highest probability of comfort point which means the occupants will be
uncomfortable but the least uncomfortable possible. These points are often the
transition points between preference variations. In other words, the algorithm may not
find comfortable temperatures when comfort preferences are significantly varying.
Table 1 summarizes the changes over time and the data communicated by the
subjects. The maximum, average, standard deviation, and minimum personal
variation over data points were 0.2813, 0.118, 0.0623, and 0.0125 °C per data point,
respectively. The maximum, average, standard deviation, and minimum personal
variations over time were 0.8486, 0.0606, 0.1591, and 0.0004°C per day, respectively.
© ASCE
Thee average daata points co ommunicated per day w was 4.23. Thhe major obsservation is
thatt thermal preeferences off individuals change conssiderably evven on small time scales
(i.ee., day to day).
d Consequently, perrsonal therm mal comfortt modeling techniques
sho ould considerr inherent tim
me variation
ns of preferennces as a maajor factor.
Figure
F 2. Six
x sample sub
bjects’ thermal prefereence variations over daata points
© ASCE
Table 1. Data acquisition details

Comfort
Test Number Comfort
Duration variation
Subject of data Start date* End date* variation
(days) (°C/ data
ID points (°C/ day)
point)
1 170 11/02/2014 15/03/2014 33 0.0368 0.0057
2 40 11/02/2014 02/04/2014 51 0.0764 0.0012
3 68 11/02/2014 02/04/2014 51 0.0575 0.0015
4 53 11/02/2014 09/04/2014 57 0.1672 0.0027
5 73 11/02/2014 12/03/2014 29 0.0761 0.0065
6 48 11/02/2014 4/11/2014 59 0.1714 0.0024
7 52 11/02/2014 27/03/2014 44 0.1395 0.0037
8 39 24/02/2014 06/03/2014 10 0.1714 0.0667
9 202 27/02/2014 3/20/2014 21 0.0714 0.0327
10 106 03/03/2014 10/04/2014 38 0.1254 0.0092
11 123 03/03/2014 23/03/2014 20 0.0913 0.0283
12 137 15/03/2014 30/03/2014 15 0.1479 0.0904
13 105 24/03/2014 07/04/2014 14 0.1854 0.0990
14 101 24/03/2014 07/04/2014 14 0.0654 0.0339
15 96 28/03/2014 17/04/2014 20 0.1165 0.0278
16 43 30/05/2014 23/06/2014 24 0.0919 0.0068
17 35 20/06/2014 25/06/2014 5 0.0904 0.1267
18 45 15/06/2014 31/07/2014 46 0.0227 0.0004
19 65 15/06/2014 31/07/2014 46 0.1350 0.0041
20 48 26/03/2014 16/04/2014 21 0.1107 0.0119
21 36 11/03/2014 24/03/2014 13 0.1363 0.0291
22 46 24/03/2014 11/04/2014 18 0.1053 0.0148
23 102 24/03/2014 30/03/2014 6 0.1531 0.4335
24 63 27/03/2014 03/04/2014 7 0.0125 0.0162
25 48 22/03/2014 03/04/2014 12 0.1853 0.0616
26 96 07/04/2014 12/04/2014 5 0.2212 0.8486
27 53 15/10/2012 15/11/2012 32 0.1620 0.0084
28 36 15/10/2012 15/11/2012 32 0.1842 0.0064
29 120 15/10/2012 01/12/2013 90 0.0291 0.0004
30 45 15/10/2012 15/11/2012 32 0.1423 0.0062
31 27 15/10/2012 15/11/2012 32 0.0164 0.0004
32 53 15/10/2012 15/11/2012 32 0.1143 0.0060
33 19 15/10/2012 15/11/2012 32 0.2813 0.0052
*Date format: DD/MM/YYYY
DISCUSSION AND CONCLUSIONS
In this paper, we demonstrated how personal thermal comfort varies over time
through studying thermal preferences of 33 subjects. We briefly described the
adaptive stochastic modeling technique that was used to quantify personal thermal
comfort. Our stochastic models are probability distributions in a dynamic Bayesian
network that utilizes a sliding window based algorithm for detecting significant
statistical differences in joint probability distributions. By applying the requirements
for standard ASHRAE 55 to the approach, we calculated comfortable temperature
ranges for each individual as they vary over time. We then calculated the absolute
© ASCE
difference of comfortable temperature ranges to the previous data point and day. Our
results suggest that personal preferences have considerable variations over time and
thus are not negligible. The average variation was 0.0606 °C with a high standard
deviation of 0.1591 °C. This finding not only shows that personal comfort should be
tracked over time (time is not defined explicitly), but also suggests that comfort
variations vary from person to person.
ACKNOWLEDGEMENTS

Foundation under Grant No. 1351701. Any opinions, findings, and conclusions or
recommendations expressed in this material are those of the authors and do not
necessarily reflect the views of the National Science Foundation.
REFERENCES
Book, Buildings Energy Data. 2010.US Department of Energy, 2011.

Frontczak, Monika and Pawel Wargocki. 2011. "Literature Survey on how Different
Factors Influence Human Comfort in Indoor Environments." Building and
Environment 46 (4): 922-937.
Ghahramani, Ali, Farrokh Jazizadeh, and Burcin Becerik-Gerber. 2014. "A
Knowledge Based Approach for Selecting Energy-Aware and Comfort-Driven
HVAC Temperature Set Points." Energy and Buildings 85: 536-548.
Ghahramani, Ali, Chao Tang, and Burcin Becerik-Gerber. 2015. "An Online
Learning Approach for Quantifying Personalized Thermal Comfort Via Adaptive
Stochastic Modeling." Building and Environment.
Hoyt, Tyler, Edward Arens, and Hui Zhang. 2014. "Extending Air Temperature
Setpoints: Simulated Energy Savings and Design Considerations for New and
Retrofit Buildings." Building and Environment.
James, Frederick. 2006. Statistical Methods in Experimental Physics. Vol. 7 World
Scientific Singapore.
Jendritzky, Gerd and Richard de Dear. 2009. "Adaptation and Thermal
Environment." In Biometeorology for Adaptation to Climate Variability and
Change, 9-32: Springer.
Liu, Weiwei, Zhiwei Lian, and Bo Zhao. 2007. "A Neural Network Evaluation Model
for Individual Thermal Comfort." Energy and Buildings 39 (10): 1115-1122.
Nicol, J. Fergus and Michael A. Humphreys. 2002. "Adaptive Thermal Comfort and
Sustainable Thermal Standards for Buildings." Energy and Buildings 34 (6):
563-572.
Peel, Murray C., Brian L. Finlayson, and Thomas A. McMahon. 2007. "Updated
World Map of the Köppen-Geiger Climate Classification." Hydrology and Earth
System Sciences Discussions 4 (2): 439-473.
Standard, ASHRAE. 2004. "Standard 55-2004." Thermal Environmental Conditions
for Human Occupancy.
U.S. Energy Information Administration. 2011. Emissions of Greenhouse Gases in
the United States 2009, Distribution of Total U.S. Greenhouse Gas Emissions by
End-use Sector, Table 3: U.S. Energy Information Administration;.
© ASCE
Yang, Zheng and Burcin Becerik-Gerber. 2014. "The Coupled Effects of

Personalized Occupancy Profile Based HVAC Schedules and Room
Reassignment on Building Energy use." Energy and Buildings 78: 113-122.
Yang, Zheng and Burcin Becerik-Gerber. 2015. "A Model Calibration Framework for
Simultaneous Multi-Level Building Energy Simulation." Applied Energy 149:
415-431.
Yao, Runming, Baizhan Li, and Jing Liu. 2009. "A Theoretical Adaptive Model of
Thermal comfort–Adaptive Predicted Mean Vote (aPMV)." Building and
Environment 44 (10): 2089-2096.
© ASCE
Occupant Individual Thermal Comfort Data Analysis in an Office
Jie Zhao1; Khee Poh Lam1; Vivian Loftness1; and B. Erik Ydstie2
1
Center for Building Performance and Diagnostics, School of Architecture, Carnegie
Mellon University, Pittsburgh, PA 15213. E-mail: jayzhao@cmu.edu
2
Department of Chemical Engineering, College of Engineering, Carnegie Mellon
University, Pittsburgh, PA 15213. E-mail: ydstie@cmu.edu
Abstract
Occupant thermal comfort is a key performance metric for green buildings.
Although numerous studies have used thermal comfort as a criterion to control
mechanical systems and optimize their energy consumptions, few studies examined
individual thermal comfort preferences and their implications. This study aims to
investigate the individual thermal comfort preferences in an office space. An online
dashboard is developed to collect subjective thermal comfort data from office
workers. The data then is compared and correlated with the measured key
environmental data, such as air temperature and relative humidity. The data analysis
results suggested that different occupants have various preferences about thermal
comfort yet each individual has fair consistent thermal preference over time. In future
study, the dashboard could be linked to the real-time control system to provide
individual thermal control for office workers.
INTRODUCTION
Providing thermal comfort to occupants is one of the most critical objectives
of green buildings. Predicted mean vote (PMV) which can be calculated by six major
factors – air temperature, relative humidity, mean radiant temperature, air velocity,
metabolic rate, and clothing insulation, are typically used as a metric to quantify
occupant thermal comfort (ASHRAE, 2010). A number of studies have used PMV as
a control criterion in the heating, ventilation, and air conditioning (HVAC) system
(Daum, Haldi, & Morel, 2011; Murakami, Terano, Mizutani, Harada, & Kuno, 2007;
Yang & Su, 1997; Yu, Loftness, & Yu, 2013).
PMV could be used to define the average thermal comfort level in a group
using calculation methods, but could also be used as a subjective metric to evaluate
the thermal sensation of each individual. Several studies have indicated that
occupants do not have uniformed thermal sensation in the same thermally controlled
space (Klein et al., 2012; Park, 2013). In order to meet the thermal comfort
requirements of different occupants, several studies have developed individual
thermal comfort voting systems to feedback the subjective thermal preference
information into HVAC control systems (Daum et al., 2011; Klein et al., 2012; Lee &
Malkawi, 2014; Murakami et al., 2007). Choi (2010) developed a wearable bio-
sensing adaptive control system that can collect thermal measurement data from
© ASCE
ind
dividual bio-ssensors to auutomaticallyy adjust HVA AC system ssetpoints (C Choi, 2010).
Thee experimen nt and simulaation resultss of these preevious studiies show thaat providing
occcupants witth individuaal thermal controls nnot only caan improvee occupant
satiisfaction (Chhoi, 2010; Murakami
M ett al., 2007) aand comfortt (Choi, 2010; Daum et
al., 2011; Kleiin et al., 20012; Lee & Malkawi, 22014), but aalso impact the HVAC
eneergy consum mption (Cho oi, 2010; Daaum et al., 2011; Kleinn et al., 20012; Lee &
Maalkawi, 2014; Murakami et al., 2007)).
udy aims to investigate the individuual thermal comfort preeferences in
This stu
an office spacce. An onlin ne dashboarrd is develooped to colllect subjectiive thermal
commfort data frrom office workers.
w Thee data then iss compared and correlatted with the
measured key environmenttal data, succh as air tem mperature annd relative hhumidity, in
ord
der to identify
fy the thermaal comfort prreference diffferences of the office w
workers.
EX
XPERIMEN
NT SETUP
A 2-story, 2,262m m2 office building - Phipps C Center for Sustainable

Lanndscapes (CSL) located in the subu urban area off Pittsburgh,, Pennsylvannia, USA is
cho
osen as a testt-bed buildinng to collectt occupant suubjective theermal comfoort feedback..
A total
t of 15 office
o workerrs in the opeen office spaaces of the bbuilding parrticipated in
the study volu untarily for three
t monthhs from Octtober to Deccember 2013. The key
env
vironmental data, such as a air tempeerature and rrelative hum midity, are coollected via
the facility’s bu
uilding autommation systeem (BAS).
Figure 1. Online occcupant subjective feedb

back data coollection sysstem.
Figure 1 shows a screenshott of the onnline dashbooard. The iinterface is

dessigned usingg HTML an nd PHP to collect
c occuupants’ indivvidual therm
mal comfort
level with the question
q of “How
“ are yo
ou feeling (thhermally)?” and clothingg insulation
© ASCE
factor with the question of “What are you wearing?” The thermal comfort band
ranges from -2 to 2 based on the PMV definition. The clothing insulation band ranges
from 0.4 to 1.9, representing from “shorts & T-shirts” to “jacket and overcoat”. The
occupants are required to provide the two pieces of information concurrently when
they are not feeling thermally comfortable. All the data are stored in a web-based
MySQL database with timestamps.
RESULTS
In order to find useful patterns of occupant thermal comfort preference, the

subjective thermal comfort votes from the web-based database and the objective
thermal environment measurement data from the test-bed building BAS system are
mapped based on spatial and temporal information. Among the 15 volunteers, only
three occupants (20%) have more than 30 effective voted instances on both thermal
comfort and clothing insulation concurrently, as shown in Figure 2. Eight (53%)
occupants have less than 30 effective voted instances and four occupants (27%) have
no effective voted instance over the experiment period. From the statistical point of
view, three datasets with more than 30 effective voted instances are mapped with
their corresponding thermal zone air temperature and relative humidity measured data,
respectively.
70
60
Comfort vote counts
50
40
30
20
10
0
A B C D E F G H I J K L M N O
Participated occupant
Figure 2. Statistics of occupants’ effective voted instances during the experiment

period.
Figure 3 - Figure 5 illustrate Subject A’s marginal plots of air temperature,

relative humidity, and clothing insulation (y-axis) versus the thermal comfort vote (x-
axis), respectively. The histograms (N = 60) of the variables are shown on the top and
right sides of the plot areas. In general, Subject A feels cold for the majority of votes
regardless of actual air temperature (20 - 25°C) and relative humidity (20 - 60%)
measurements. One noticeable pattern is that all the “very cold” votes are the times
when the air temperature drops below 21°C. Figure 5 shows that most of the “feeling
cold” votes are correlated with clothing insulation factor above 1.5, which suggests
that although Subject A has adapted his/her clothes based on the thermal comfort
feeling, he/she was still feeling cold during the study period.
© ASCE
Marginal Plot of Air temperature (°C) vs Comfort vote (PMV)
26 Feeling cold
regardless of
Air temperature (°C)

24 temperature
22
20
-2 -1 0 1 2
Comfort vote (PMV)
Figure 3. Marginal plot of air temperature vs. comfort vote of Subject A.
Marginal Plot of Re lative humidity (%) vs Comfort vote (PMV)
Feeling cold
60
Relative humidity (%)
regardless of
humidit y
50
40
30
20
-2 -1 0 1 2
Comfort vote (PMV)
Figure 4. Marginal plot of relative humidity vs. comfort vote of Subject A.
Marginal Plot of Clothing insulation (Clo) vs Comfort vote (PMV)
2.0
Clothing insulation (Clo)
Wearing thick clothes

but still feeling cold
1.5
1.0
0.5
-2 -1 0 1 2
Comfort vote (PMV)
Figure 5. Marginal plot of clothing insulation vs. comfort vote of Subject A.
Figure 6 - Figure 8 show Subject B’s marginal plots of air temperature,

relative humidity, and clothing insulation (y-axis) versus the comfort vote (x-axis),
© ASCE
respectively. The histograms (N = 36) of the variables are shown on the top and right
sides of the plot areas. In general, Subject B feels cold for the majority of votes
regardless of actual air temperature (20 - 25°C) and relative humidity (20 - 60%)
measurements. Figure 8 shows that almost all the “feeling cold” votes (except for one
vote) are correlated with clothing insulation factor around 0.5, which suggests that
Subject B has relatively thin clothes in the office compared to the typical business
dress. Arguably, Subject B could have improved his/her thermal comfort level easily
by adding a jacket or another layer of clothes. This result is very different from the
data of Subject A, who has already had thick clothes but still feels cold.
Marginal Plot of Air te mperature (°C) vs Comfort vote (PMV)
26 Feeling cold
regardless of
temperature
24
22
20
-2 -1 0 1 2
Comfort vote (PMV)
Figure 6. Marginal plot of air temperature vs. comfort vote of Subject B.
Marginal Plot of Relative humidity (%) vs Comfort vote (PMV)
60
Feeling cold
Relative humidity (%)
regardless of
50 humidity
40
30
20
-2 -1 0 1 2
Comfort vote (PMV)
Figure 7. Marginal plot of relative humidity vs. comfort vote of Subject B.
© ASCE
2.0

1.5
Consistently wearing thin clothes
and feeling slightly cold
1.0
0.5
-2 -1 0 1 2
Comfort vote (PMV)
Figure 8. Marginal plot of clothing insulation vs. comfort vote of Subject B.
Figure 9 - Figure 11 show Subject C’s marginal plots of air temperature,

relative humidity, and clothing insulation (y-axis) versus the comfort vote (x-axis),
respectively. The histograms (N = 41) of the variables are shown on the top and right
sides of the plot areas. In general, Subject C feels warm, especially when the air
temperature is above 22°C and relative humidity is above 45%. Figure 8 shows the
clothing insulation factor is mostly below 1.0, which suggests that Subject C has
relatively thin clothes in the office compared to typical business dress, but the subject
still feels warm in general. This is clearly different from the other two subjects’
profiles.
Marginal Plot of Air tempe rature (°C) vs Comfort vote (PMV)
26
24
22
Feeling warm
20
in general
-2 -1 0 1 2
Comfort vote (PMV)
Figure 9. Marginal plot of air temperature vs. comfort vote of Subject C.
© ASCE
Marginal Plot of Relative humidity (%) vs Comfort vote (PMV)
60
Relative humidity (% )
50
40
30
20 Feeling warm
in general
10
-2 -1 0 1 2
Comfort vote (PMV)
Figure 10. Marginal plot of relative humidity vs. comfort vote of Subject C.
2.0
1.5
1.0
Wearing less clothes
but still feeling warm
0.5
-2 -1 0 1 2
Comfort vote (PMV)
Figure 11. Marginal plot of clothing insulation vs. comfort vote of Subject C.
By comparing the three subjects’ data and taking into consideration of the
overall voting statistics, the following conclusions can be drawn. A total of 11 out of
15 volunteers express their opinions on their thermal comforts when they do not feel
comfortable, which indicates thermal comfort is something important for the test
group to consider at workplaces. However, people have very different sensitivities to
thermal comfort. Three subjects voted over 30 times during the three months
experiment periods, but most people voted less than 30 times. In addition, among the
three subjects, the data shows very different comfort preferences and clothing choices.
However, the thermal comfort preference and clothing choice of each individual are
mostly consistent for the majority of time during the entire experiment period.
Therefore, it is necessary to develop a control system that can accommodate
individual occupant thermal comfort preference, and based on the individual
preference, the control system could make consistent adjustment to meet individual
needs.
DISCUSSIONS
This study found very interesting patterns about individual thermal comfort
preference despite the small sample size (15 people) and the relative short duration
(three months). One could imagine that if future studies could incorporate more
© ASCE
individuals under different seasons and locations that could also represent culturally
behavioral differences, the findings would be very valuable for understanding the
occupant thermal comfort and behavior, and to develop more appropriate occupant-
oriented control systems.
Future work could be focused on developing a control system that linking the
online dashboard into the actual HVAC zone level control, and testing its comfort and
energy performance.
CONCLUSIONS
This study investigated the individual thermal comfort preference in an office

space. An online dashboard was developed to collect subjective thermal comfort data
from office workers in a test-bed building. Key environmental data, such as air
temperature and relative humidity were collected through the BAS system during the
same period. The data analysis results showed that thermal comfort is an important
factor that occupants are aware of in their workplaces. Beyond this awareness, it is
evident that individuals in the test group have very different sensitivities to thermal
comfort, and among the most sensitive individuals, there are different comfort
preferences and clothing choices. However, for each individual, thermal comfort
preference and clothing choice remain consistent for the majority of time during the
experiment period. These findings strongly support the need for providing
individualized thermal comfort control systems for office workers.
ACKNOWLEDGEMENTS
The authors would like to acknowledge National Science Foundation (NSF)

Emerging Frontiers in Research and Innovation (EFRI) in Science in Energy and
Environmental Design (SEED) (Award #: 1038139) and Phipps Conservatory and
Botanical Gardens for funding the study. The authors would also like to acknowledge
Bertrand Lasternas, Ray Yun, Haopeng Wang, Omer Karaguzel, and Xuan Luo from
Carnegie Mellon University Center for Building Performance and Diagnostics for
their contributions to this study.
REFERENCES
ASHRAE. (2010). ASHRAE Standard 55-2010: Thermal Environmental Conditions

for Human Comfort. Atlanta: American Society of Heating, Refrigerating and
Air-Conditioning Engieers, Inc.
Choi, Joon Ho. (2010). CoBi: Bio-Sensing Building Mechanical System Controls for
Sustainably Enhancing Individual Thermal Comfort (PhD Dissertation).
Pittsburgh: Carnegie Mellon University.
Daum, David, Haldi, Frédéric, & Morel, Nicolas. (2011). A personalized measure of
thermal comfort for building controls. Building and Environment, 46(1), 3-11.
Klein, Laura, Kwak, Jun-young, Kavulya, Geoffrey, Jazizadeh, Farrokh, Becerik-
Gerber, Burcin, Varakantham, Pradeep, & Tambe, Milind. (2012).
© ASCE
Coordinating occupant behavior for building energy and comfort management

using multi-agent systems. Automation in Construction, 22, 525-536.
Lee, Yoon Soo, & Malkawi, Ali M. (2014). Simulating multiple occupant behaviors
in buildings: An agent-based modeling approach. Energy and Buildings, 69,
407-416.
Murakami, Yoshifumi, Terano, Masaaki, Mizutani, Kana, Harada, Masayuki, &
Kuno, Satoru. (2007). Field experiments on energy consumption and thermal
comfort in the office environment controlled by occupants’ requirements from
PC terminal. Building and Environment, 42, 4022-4027.
Park, Jihyun. (2013). Post-occupancy Evaluation for Energy Conservation, Superior
IEQ & Increased Occupant Satisfaction. Paper presented at the World
Workplace 2013.
Yang, K. H., & Su, C. H. (1997). An Approach to Building Energy Savings Using the
PMV Index. Building and Environment, 32(1), 25-30.
Yu, Yuebin, Loftness, Vivian, & Yu, Daihong. (2013). Multi-structural fast nonlinear
model-based predictive control of a hydronic heating system. Building and
Environment, 69(0), 131-148.
© ASCE
Facial Skin Temperature as a Proactive Variable in a Building Thermal

Comfort Control System
Bo Yi1 and Joon-Ho Choi2

1
Building Science, School of Architecture, University of Southern California, 850
West 37th St., Los Angeles, CA 90089. E-mail: yib@usc.edu
2
Building Science, School of Architecture, University of Southern California, 850
West 37th St., Los Angeles, CA 90089 (corresponding author). E-mail:
joonhoc2@gmail.com
Abstract
The human body has a biological thermoregulation mechanism (homeostasis),

which enables it to maintain a stable and constant body temperature via changing
physiological signals such as skin temperature and heart rate. These signal patterns,
that are generated based on the human autonomic nervous system, have been
validated as a potential variable to provide information about an individual’s current
thermal sensations. Among the numerous body segments and parts generating skin
temperature, facial skin is five times more sensitive to ambient thermal conditions
than other skin surfaces. Therefore, based on the use of facial skin temperatures, this
research will establish an adaptive thermal sensation model which will be applicable
to automatic (individual) building mechanical system controls within the principle of
human-building interactive strategy.
INTRODUCTION
Thermal comfort is a condition of the mind that experiences satisfaction with

the thermal environment and is assessed by subjective evaluation (ASHRAE, 2013).
According to the ASHRAE standard, the six primary factors that affect thermal
comfort can be categorized into two groups: personal factors (due to characteristics of
a building’s occupants) and environmental factors (due to ambient conditions).
Personal factors include clothing insulation and metabolic rate; environmental factors
include air temperature, relative humidity, air speed, and mean radiant temperature
(ASHRAE, 2013). These factors can affect thermal comfort since they determine the
extent of heat gain and loss of the human body. Thermal comfort is a condition of
mind that indicates how people feel about the thermal condition of their ambient
environment. Achieving thermal comfort of the occupants of a building is one of the
main tasks of HVAC (heating, ventilation, and air conditioning) systems, while
keeping energy use for HVAC at a minimum. Satisfaction with the thermal
environment is significant because of its influence on the productivity and health of
© ASCE
the individuals involved. After

A receivin ng 34,000 suurvey responnses to therm
mal comfort
queestions in 21
15 buildings, researcherss from UC B Berkeley fouund that offi
fice workers
who are satisfieed with theirr thermal environment aare more prooductive thann those who
werre dissatisfiied (Huizen nga, Abbaszzadeh, Zaggreus, & A Arens, 2006)). Thermal
disccomfort hass also been n identified as contribuuting to sicck buildingg syndrome
sym
mptoms (My yhren & Holmberg, 200 08). On the other hand,, thermal discomfort is
signnificantly afffected by individual
i physiologica
p al and psychhological m mechanisms.
Disscomfort is linked to thermal
t streess, which ccan affect w work perforrmance and
ind
dividual heallth (Wyon, 1996).
1 According to H Huizenga’s reesearch, 80 % or more
occcupants claimmed that theey were satissfied with thhermal condditions (Huizzenga et al.,
20006).
Currenttly, most bu uilding therm mal environnmental conntrols and ssystems are
adoopting prediccted mean votev / predictted percentaage of dissattisfied (PMV
V and PPD)
models by utillizing heat balance
b equaations to esttimate therm
mal comfortt conditions
thatt affect a buiilding’s tenaants. Fanger’’s equations are used forr PMV calcuulations of a
larg
ge number of human samples with a particuular thermal condition.. This is a
commbination off dry ball aiir temperatu ure, mean raadiant tempeerature (MR RT), relative
hummidity, air sppeed, metabo olic rate, and
d clothing innsulation (KE
E, 2003) (Figgure 1).
Figure 1. Psychometri
P ic chart-theermal comfoort range off the PMV M Method.
Howeveer, a lot of building
b occcupants repoort their therrmal stress, discomfort,
andd dissatisfacttion during their time in a built ennvironment, in spite of tthe thermal
connditions mon nitored and regulated by y the currennt Fanger’s model. IFM MA reported
thatt too cold and
a too warrm condition ns were thee most criticcal issues afffecting the
occcupants’ ind door environnmental quaality, includding lightingg, spatial, pprivacy, air
quaality, etc. While
W many efforts havee been madee to overcom me the currrent control
appproaches thaat rely on conventional model-baased enviroonmental coontrols, the
therrmal comforrt issue has not
n been reso olved. This iis a critical llimitation inn the current
buiilding enviroonmental co ontrol strateg
gies, and it is essential that a hum man-building
inteegrative fram
mework be developed to t enhance human phyysiological bbenefits and
envvironmental sustainabilitty via optimiization of ennergy use.
© ASCE
The human body has a biological thermoregulation mechanism (homeostasis)

that enables it to maintain a stable and constant body temperature via changing
physiological signals such as skin temperature and heart rate. These signal patterns,
that are generated based on the human autonomic nervous system, have been
validated as a potential variable for providing information about an individual’s
current thermal sensations. Among the numerous body segments and parts generating
skin temperatures, facial skin is five times more sensitive to ambient thermal
conditions than other skin surfaces. Therefore, based on the use of facial skin
temperatures, this research will establish an adaptive thermal sensation model that
can be applicable to automatic (individual) building mechanical system controls
within the principle of human-building interactive strategy.
This study conducted a series of experiments with human subjects in an

environmental chamber by collecting each individual’s facial skin temperature and
thermal sensation in real time while ambient thermal conditions were being changed.
The collected subjective and objective data were processed using multiple data
mining tools, such as a decision tree, neural network, and clustering, to develop a
facial-skin temperature driven thermal sensation model. The developed model was
also validated using human subjects who had not participated in the previous tests in
order to prevent any over-fitting effect of the model.
The outcome of this research, in the form of a computational model that uses
real time facial skin temperature data as an input variable, will be applicable to any
existing thermal control system in a building. This especially applies to any
individual control system that is equipped with a personal environmental module /
terminal reheating box, and is situated in an office building or a healthcare facility.
The occupants’ low mobility in these facilities can be helpful for collecting the facial
skin temperature data remotely without being intrusive.
METHODS
This study conducted a series of experiments with human subjects in an

environmental chamber. Since the study focused on a workplace environment, we set
up a workstation in the chamber and test participants were asked to generate light
office work, such as typing and a web search in each test. In this study, 15 subjects
participated in the test, and each test was conducted for 100 minutes, which included
the times for wearing sensors and waiting in a standby condition.
During the experiment, the temperature was controlled within a range of 20ºC
to 30ºC. The sequence began from cooling to heating, or vice versa, to prevent any
biased thermal sensation reporting from any test participant. Most human subjects
were either undergraduate or graduate students at the University of Southern
California. We selected 1.5ºC as a temperature change step based on the capacity of
the HVAC systems in the environmental chamber. Each test consisted of seven
change steps and, at the end of each step, the subject was asked to report his/her
© ASCE
therrmal sensatiion and com

mfort conditio
on using a 77-point scalee, as illustratted in Table
1 below.
Figure
F 2. Exxperiment prrocedure
A 7-point scale is a very pop pular methood for surveeying user ssatisfaction;
ASHRAE-55 also a adopted this scale to o estimate th ermal comfoort and sensaation. Since
it is
i neither to oo complicaated nor too o simple, a 7-point scaale has beenn popularly
adoopted in the research do omain of ind door environnmental quaality and envvironmental
satiisfaction stu
udies. In ord
der to obtainn stable expeerimental coonditions, suubjects were
req
quested not tot have food d for at leastt 30 minutess before the experiment in order to
maintain consisstent metabo olic rates of the
t individuaal subjects.
Figure 3. An experimenta
e al setting in
n this
study Figure 4. Selected faacial
temperaturre sensing ppoints
RE
ESULTS
Individu
ually collectted skin tem
mperatures aat six facial points of each subject
werre categorizeed based on the thermal sensation annd conditionning mode, ii.e., cooling
andd heating. Figure 6 illlustrates thee generated temperaturees (includinng dry bulb
tem
mperature annd operative temperaturee), and the aaverage skinn temperaturre collected
from
m one samp pled subject during
d xperiment. IIt looks veryy stable in Fiigure 6, but
the ex
© ASCE
the average skiin temperatu ures varied depending

d oon the test paarticipants. IIn addition,
eveen though thee thermal co
onditions gen nerated in thhe chamber w were almost the same or
verry similar, th
he subjects’ reported theermal sensattions were tootally differeent because
of their
t physiological or peersonal cond ditions. How wever, this wwas very obvvious when
commpared with h the existingg literature. As shown in Figure 7,, two samplled subjects
reported entirelly different sensations, even thoughh the same thermal proocedure was
useed during thee test.
Tab
ble 1. Therm
mal sensatio
on and comffort survey
1. What
W is your overall
o level of
o thermal comfort?
Very Unsatisfied Slightly Neural Slightly Satisfied Very
unssatisfied unsatisfied satisfied satisfied
□ □ □ □ □ □ □
2. What
W is your overall
o therm
mal sensation??
Very cool Slightly Neural Slightly warm Very
Cool cool warm warm
□ □ □ □ □ □ □
Fiigure 5. Tim
me series plo
ot of Tdry (d
dry bulb airr temperatu
ure) and To ((operative
tem
mperature) and T skin (average)
Figure 6. Thermal sensation reported

r by two differeent test subjects
© ASCE
Figure 8 also illusttrates how diverse

d tempperatures weere measureed from six
facial skin poinnts and the average
a of on
ne sampled subject. Thiis finding inndicates that
ind
dividual senssing points ono a face gen nerate differeent levels off skin tempeerature even
though those sensing spo ots are exp posed to thhe same theermal condditions. Not
surpprisingly, all of the test subjects sho
owed dynam mic patterns pper sensing ppoint, and it
wass nearly imp possible to fiind a consisttent rule for the measureed skin tempperatures of
all subjects.
Based on
o this findin ng, the averaage facial skkin temperatture was seleected as the
representative variable
v for a thermal inndex of a facce. The data from all of tthe subjects
werre assembled d and categoorized into seven
s differeent thermal ssensations, aas shown in
Fig
gure 9. The absolute
a valuues of the tem
mperatures w were totally different, deepending on
the sensing sp pots on a face,
f and allso on the test subjectts. Their phhysiological
therrmoregulatioon principlees were very y similar to each other in order to maintain a
heaat balance beetween the human
h body and the ambbient thermaal condition.. Therefore,
5 minutes
m wass selected ass a time wiindow framee for estimaating a graddient of the
aveerage facial skin temperrature per th hermal sensaation while the temperaature in the
chaamber was co ontinuously being chang ged.
Figure 7. Pattterns of skiin temperatures at six sselected faciial skin poin nts, as air
tempeerature (fro om one samp pled subjectt).
Figure 9 summariizes the diistribution oof the graddient of avverage skin
mperature perr thermal sen
tem nsation, baseed on the collected data from all of tthe subjects..
ANNOVA (Anaalysis of variance) tesst results sshowed signnificant diffferences in
graadients betw
ween any two o different thermal
t sennsations. To increase the statistical
signnificance, th
he cold sennsation data (i.e., -3, -22. -1) weree grouped aas one cool
sen
nsation, and the
t warm sen nsations (i.ee., +3, +2. +11) were also combined aas one warm
sen
nsation, while keeping the neutrall sensation as a singlee baseline nnorm. This
trannsformation of the data generated
g mo ore highly siignificant loow p-values < 0.00.
© ASCE
Figure
F 8. Ch
hange rate ranges
r of the average sk kin temperaature of all subjects'
facial skin pointts.
DIS
SCUSSION
N
The findings in thiss research leed to the connclusion thaat a gradientt of average
skinn C is a critiical indicatorr for revealin
ng each subjject’s thermaal sensation. In addition,,
eveen though the absolute average
a skin temperaturee was the samme betweenn subjects, it
cou
uld be a goo od indicatorr to illustratee an “overaall” sensationn per person since the
ran
nge of skin temperatures
t s for each sensation w was not widee enough. O On the other
han
nd, an ambieent temperatture was also o recognizedd as a signfiicant parameeter since it
wass good enough for illustrrating the “o overall” therm
mal sensatioons of the subbjects.
Figuure 9. Sensaation and ACC Task deci sion in coolling process.

Based on
o these deteerminations, study investtigators seleccted averagee facial skin
tem
mperuatre, grradient of the
t average skin tempeerature, and the ambiennt dry bulb
tem
mperature, as well as gender
g (duee to its criccial role in thermal seensation) as
© ASCE
parrameters. By y adopting these param meters and consideringg the humann factor, a
deccision tree (u
using J48) was
w establish hed per the H HVAC systeem model (ii.e., cooling
and
d heating). Figures 9 and d 10 illustratee how the deecision trees was appliedd to AC and
heaater controlss. Depending g on the esttimated therrmal sensatiion, the conndenser and
heaater were “onn” or “off” to
o maintain a subject’s thhermal comffort conditionn, a thermal
neu
utral sensatioon. For thesse decision trees,
t a 10-ffold cross vvaldiation was adopted.
Thiis estimationn approach achieved
a 78% , 86%, annd 86% accuuracy in esttimating the
therrmal sensatiions: cool, neutral,
n and warm
w condittions, respecctively, as a function of
facial skin temmperature an nd the ambieent thermal condition, aas well as ggender as a
humman factor.
Figuree 10. Sensation and Hea

ater task deecision in heeating proceess.
CO
ONCLUSION
This stu
udy focused on determin ning the poteential use off facial skin ttemperature
to estimate
e an individual’s
i thermal sennsation (espeecially in a wworkstation ssetting) and
to identify
i any
y significant responses of o facial skiin temperatuure to ambieent thermal
connditions. Forr purposes ofo testing thee relationshipp between ffacial skin teemperatures
andd thermal co onditions, an nd to identiffy potential effects, a seeries of hum man subject
expperiments weere conducteed in an env vironmental chamber loocated at thee University
of Southern
S Caalifornia. Ev
ven though absolute
a leveels of facial skin temperature were
totaally differen
nt, depending on the six x selected ssensing spotts and also the various
sub
bjects, the grradient of thee average faccial skin tem
mperature waas practicallyy consistent
for all test subbjects. In add dition, the average
a skinn temperaturre and the aambient dry
bullb temperatu ure did not reeadily reveaal the individdual thermall sensations, although it
wass possible too provide an n overall therrmal sensati on within ceertain rangess. Based on
these findingss, this stud dy adopted ambient ttemperature,, average facial skin
tem
mperature, an nd its gradieent to develoop a thermall sensation pprediction m model in the
formm of a decision tree (J4 48), with an n estimated pprediction aaccuracy of 881% ± 5%,
deppending on th he thermal sensation.
s
© ASCE
In spite of the significance of these research findings, a few research limitations

affecting this study certainly warrant further investigation. The outcome could have
been influenced by the number of test participants. The total number of participants
was 20, which is good enough for t-statistics. However, to attain robust statistical
significance in this study, sample sizes should have been larger in order to fully
validate research discoveries with regard to estimated thermal sensations as a
function of facial skin temperature. In addition, due to the significance of human
factors, such as gender, age, and body mass index, which contribute to overall
thermal sensations, a future study should consider additional human physiological
characteristics of the subject samples.
REFERENCES
ASHRAE. (2013). ASHRAE Standard 55. American Society for Heating, Ventilating
and Air- …. American Society of Heating, Refrigerating, and Air-
Conditioning Engineers.
Huizenga, C., Abbaszadeh, S., Zagreus, L., & Arens, E. A. (2006). Air quality and
thermal comfort in office buildings: Results of a large indoor environmental
quality survey. Proceeding of Healthy Buildings 2006, 3.
KE, C. (2003). Fanger’s Thermal Comfort and Draught Models (IRC-RR-162).
Myhren, J. A., & Holmberg, S. (2008). Flow patterns and thermal comfort in a room
with panel, floor and wall heating. Energy and Buildings, 40(4), 524–536.
Thermal comfort : analysis and applications in environmental engineering / by P.O.
Fanger. - Version details - Trove. (1972).
Wyon, D. P. (1996). Individual microclimate control: required range, probable
benefits and current feasibility. In International Conference of Indoor Air
Quality and Climate (pp. 1067–1072).
© ASCE
Advancing Occupant-Centered Performance Simulation Metrics

Linking Commercial Environmental Quality to Health, Behavior, and
Productivity
M. Azarbayjani¹; D. Brentrup¹; and R. Cox²
¹School of Architecture, College of Arts +Architecture, University of North Carolina

Charlotte.
²Department of Electrical and Computer Engineering; College of Engineering,
University of North Carolina Charlotte.
Abstract
Integration of the human dimension in product and process modeling, as

related to the integrated design process, requires the accounting of the impact of the
factors that determine Indoor Environmental Quality (IEQ). Their influences upon
occupant comfort remain the primary measures for predicting productivity and
satisfaction within the workplace. The aim of this proposal is to present the outcomes
of a yearlong literature review summary of four IEQ elements that most influence
occupant health, comfort and wellbeing (i.e., thermal, ventilative, luminous, and
acoustic comfort). The study’s results presented provide evidence to modify office
environmental design guidelines to improve economic, health, and environmental
benefits. Our work postulates that case study data in co-relational analysis is critical
to the development of new metrics for simulation predictability in new and
repurposed design.
Our assessment concludes that two critical steps are necessary to develop
predictive measures to insure comfort sufficiency in new and especially retrofit
buildings: 1) Because occupants are exposed to multiple factors simultaneously we
speculate that the weighted value of the interrelationships of these influences are the
critically important factor required to develop articulate sensitivity in simulation
schedules; and 2) Because most of the decisions regarding economic investments
(especially in repurposing and retrofit design) continue to be made based on the first
cost systems replacement rather than on a cohesive model of comfort sufficiency that
inclusively embraces both energy efficiency and attaining a high level of
environmental quality.
BACKGROUND
Numerous case studies have been performed over the years with respect to
Indoor Environmental Quality (IEQ). These case studies have articulated the
standards that determine the qualitative aspects of IEQ. Moreover, they have had a
marked impact on directing various design guidelines such as LEED and the current
WELL Building Standard. This study has conducted an extensive review of case
study literature on office building environments in the U.S. and abroad. It provides a
© ASCE
threshold for the reconsideration of how the indoor environmental conditions of air
quality, thermal quality, and lighting and acoustics have been evaluated and how user
satisfaction with respect to environmental quality has been differentiated. The goal of
this project has been to focus the research of the past decade. It has collated the
results of these case studies to develop an understanding of how the research is
typically conducted and to determine if the reported results contain information that
supports the development appropriate metrics for measuring commercial office
environment performance by quantifying IEQ influences on behavior.
METHOD
The primary research has involved a deep literature review, as evidenced in

the bibliography. The project explored the various case studies that have been
performed over the years with respect to IEQ in commercial and institutional
facilities. During the initial phases of the project, the team compiled this data and
assembled a detailed summary. During the latter phases of the project, the team has
focused on using the data to develop quantifiable metrics equated with an integrated
sustainable practice model; considering “People (occupant wellbeing) Planet
(behavioral levels of change associated with satisfaction resulting in energy
efficiency) and Profit (predictive return on investment associated with healthy
building practices). These metrics will focus on the combined effects of the various
components of indoor environmental quality (i.e. air quality, visual comfort, etc.)
The body of this paper is composed (as was the literature review) into the four
primary areas of research that have topically defined human comfort Thermal
Comfort, Ventilative Comfort, Luminous Comfort and Acoustic Comfort. The
following sections have been written by our research assistants. They are edited
annotations from the original 35 page document that summarizes the research. These
sections represent a sample of the outcomes from which over all conclusions have
been drawn. They provide only a sample of the complete annotated summary of
findings that will be edited and published in the project final report. The bibliography
is a complete compendium of the literature reviewed.
THERMAL COMFORT
The literature demonstrates that thermal comfort has a significant impact on

health, productivity and workplace satisfaction. LEED credits for Thermal Comfort,
EQc7.1and EQc7.2 respectively, require HVAC design compliance with ASHRAE
Standard 55-2004 and require building occupant surveys to determine whether the
working environment has satisfied the thermal conditions of ASHRAE 55. If the
result of survey shows more than 20 percent of occupants are dissatisfied, implement
corrective action plan should be considered. It shows the importance of conducting
research on thermal comfort and its impact on occupants.
This assessment quantifies occupant productivity benefits, through literature
review and comparison of papers using the Well Building and ASHRAE-55
standards. In this section, the analysis and synthesis of the literature review on the
impact of thermal comfort on health, productivity and satisfaction in addition to the
© ASCE
cost benefits have been evaluated. Thirty-eight (38) papers on the issue of thermal
comfort have been studied for the period from 2001 to 2014.
AN IMPACT OF TEMPERATURE ON PERFORMANCE
Among 38 studies, 20 papers have studied impact of temperature on

productivity. There are many ways to look at the performance in the workspace.
However, overall performance, speed, accuracy, key stroke have been considered as
key factors of productivity.
The range of temperature considered in these papers is from minimum of 20'
C (68'F) to maximum of 30' C (86'F). The range of temperature considered in these
papers is from minimum of 20' C (68'F) to maximum of 30' C (86'F). It should be
noted that all papers (except 5) conducted studies based on ASHRAE-2014,
regardless of the impact of humidity (i.e., minimum 20 'C (30% RH) to maximum
28'C (60% RH). Based on the literature review increasing temperature for
temperature above 22 'C (71.6 F) at which the highest productivity has been indicated
based on Seppanen et al research at 2006, 13 papers shows decrease in performance
and 4 papers shows increase in performance (Figure 1). The amount of Decrease and
Increase in performance respectively are from 0.3 %to 6.4 %per 1' C ,Mean 2.7 %
and from 0.93% to 7.8% ,Mean 3.1%.
10
Chatzidiakou et al. (2014), -11%
5
Cui et al, (2013), 5.4% in Memory Typing
0
Increase and Decrease in Performance
Wargocki & Wyon (2013), -4% to -2% in

15 20 25 30 35 terms of Speed by Increasing per 1 c
-5 from (20 to 25)
Bako Biro et al (2012), -8%
-10 Fisk, Black &Brunner (2011), Eliminating

Temp >23, 0.23% in Winter Performance
-15 Li Lan et al (2011), 7.5%
Li Lan et al(2010), -2.3% in terms of

-20 Speed
Tanabe S. et al (2009), -1.9%

-25
Nishihara et al, (2007), 7.8% in total
-30 number of Typing in one day
Seppanen & Fisk (2006),-10%

-35
Increase of Temperature
Figure 1. Comparison of papers in impact of Temperature on Performance
VENTILATIVE COMFORT
The following research attempts to quantify occupant productivity benefits

based on ventilation flow rates based upon literature review and comparing relevant
standards. There are many articles that address the benefit of clean, ventilated air for
both occupant productivity and overall health. From the twenty-one articles reviewed;
© ASCE
starting in the late 1990’s to the present, they range in information about productivity,
health, comfort, and creativity related to ventilation rates and air quality. The research
reported in the articles relates to the ASHRAE standards that determine ventilation air
rates. Additional articles have looked beyond the standards to see if more needs to be
done in order to provide better environmental quality. There are also standards such
as LEED and WELL that focus on indoor environmental qualities and have
ventilation standards. These different standards were compared to one another to help
to understand which one is the most effective when it comes to occupants overall
productivity and health. While each standard presents a different set of criteria for
ventilation quality, they all recognize that it is important and necessary for the
building users to have good quality air in order to be comfortable in their
environment. Unlike LEED and ASHRAE, the WELL building standard does go into
more detail about the individuals health and wellbeing, such as understanding
nourishment, physical, and mental aspects of the person, while the other standards
focus more on the buildings requirements, which is important to ventilation
requirements because it is more than just providing air but effects the overall indoor
environmental quality. The articles help show that the individual’s overall comfort is
related to a lot of factors that the building has to have / accomplish in order to meet
the basic needs of the worker. Through the research, it is clear that there should be at
least a ventilation rate of 21 to 40 cfm for optimal occupant health, along with access
to both natural and mechanical ventilation, with this the performance rate of the
workers will increase up to 8% (Fisk, 2000).
A significant number of articles address the importance of ventilation for
healthier building occupants, but not as many actually understand what that ideal
ventilation rate are. ASHRAE 62.1 – 2010 states that a building needs a ventilation
rate of 20 cfm/ per person or .06 cfm/ per square foot for in a mechanically ventilated
office space. For a naturally ventilated space, there must be a minimum size of
openings based on floor area to be ventilated. Most articles follow the ASHRAE
standards as a minimum requirement for ventilating the designated space, but they
compare having a mechanically ventilated versus a mix-mode or completely natural
ventilated space. The article “Mixed Mode Ventilation HVAC meets Mother
Nature,” focuses on occupant preference for operable windows in their office space to
control their own natural ventilation. Brager, Ring, and Powell (2000) study points
out that occupants with access to an operable window are comfortable over a larger
range of temperatures than people in completely ventilated spaces.
There are many factors that affect the comfort of an occupant in a building as
related to both ventilation and air quality in a space. But, few articles discuss the
ventilation rate solely in its relationship to productivity and satisfaction. Each article
(Figure 2) used a specific ventilation method (i.e., mechanical or naturally
ventilated). Some research reported the actual measured rates used. As shown, the
spaces with natural ventilation increased in satisfaction the most, with one article
almost reaching 100% satisfaction with their environment. The least satisfaction was
reached when the ventilation rate was only 17.7cfm, as discussed earlier, this is below
the standard ASHRAE which states a good rate is at least 21cfm. An article presented
by Helsinki University states that improved ventilation rates up to 17L/s per person
improves office task performance and a rate above 10L/s to 25L/s add a significant
© ASCE
deccrease to sick
k building sy
yndrome. Ventilation noot only provvides new, cllean air, but
also
o reduces thhe amount off pollutants and airbornee viruses in a space. Thherefore, the
abo
ove rates aree an importaant factor in
n the overalll health and productivityy within an
offi
fice.
Figgure 2. Occuupant satisfacction as a facctor of Natuural and mechhanical
venttilation ratess
In conclusion 23 so ources have been

b review
wed (articles and standarrds) ranging
fromm the late 1990’s
1 to prresent. Thee research illlustrated goood progress regarding
where ventilatiion rates sho ould be to achieve
a greaater occupannt satisfactioon. Though,
therre is still a lot of informmation missiing when it comes to not only venttilation, but
oth
her factors th hat cause thee ventilation rate to needd to change ddue to air quuality, open
or closed wind dows, and th he amount of o people in or out of thhe space. O Over all, the
reseearch provides a good d basis for understandiing the impportance andd need for
venntilation in an
a office space. It helps with
w overalll productivityy, which leaads to better
emp ployee’s heaalth, and therefore more profit and pproficiency ffor any givenn company.
Acccording to an a article by y Indoor En nvironment D Department,, increasing ventilation
red
duces related d illnesses byy 18%, thereefore saving up to $14 bbillion on sicck leave per
yeaar and increaase in work performancce (Health an and Productiivity Gains ffrom Better
Ind
door Environ nments and their Relatiionship withh Building Energy Effi ficiency). A
goood next step in this reseaarch would be b to create a tool that uuses resourcees to collect
data on health and producttivity. Comp panies can thhen use this tool to help understand
the benefits and d see the inccrease in oveerall productiivity, healthh, and econommic savings
n then use thiis tool.
can
LU
UMINOUS COMFORT
C T
Light iss one of thee fundamenttal elements of architecture. It deffines space,

rev
veals form, and
a makes or o breaks an individual’ s experiencee of a buildiing. It is a
cru
ucial part bo
oth of occup pant comfortt and produuctivity. Ligght can affeect comfort,
mood, alertness, productiv vity, and eveen physical health in w
which it playys no small
© ASCE
part. It possesses both quantitative and qualitative aspects that must be considered in
design, and the role it plays in the workplace especially cannot be overstated. This
research attempts to analyze the benefits of the luminous environment on health,
productivity, and satisfaction. It begins by recognizing two different types of light
daylight and electric light. For each of these types of light, articles were found,
analyzed, and compared with existing standards (LEED and WELL) to find areas
where the standards may be lacking or find potential opportunities for further
research. Overall, 96 articles were taken into account, spanning from 1965 to 2015.
The standards set forth by LEED and WELL go beyond building codes and deal with
environmental responsibility and occupant health respectively. In general, WELL
tends to go beyond LEED in assuring that the built environment is conducive to the
well-being of occupants, with credits that deal with the circadian rhythm, glare
control, and lease depth.
THE IMPACT OF LIGHTING ON HEALTH THROUGH CIRCADIAN

RHYTHM
Light has a profound effect on health, largely through its influence on the
circadian rhythm. Certain luminance values and color temperatures tend to increase
alertness, well-being, and vitality. According to a study by K. C. H. J. Smolders, de
Kort, & van den Berg (2013) Brighter light tends to mean higher vitality in
occupants, especially during morning hours (Figure 3). Leichtfried et al. (2015)
found that exposure to bright light around 5000lx from 740 to 810am can improve the
mood of occupants, however, it can also increase distraction thus lowering mental
performance. A study by Sahin & Figueiro (2013) suggests that a 48 minute, post-
lunch exposure to red light (40lx, 630nm) can increase alertness in workers. Figures 3
and 4 illustrate the various illuminance and color values throughout the day tested
within the articles.
6000 8000
Lighting Level (lux)
Color Temperature
5000 6000
4000 4000
(Kelvins)
3000 2000
2000 0
1000
0
Time of Day Time of Day
Figure 3. Hourly change in illuminance Figure 4. Hourly change in Color temperature
The research suggests that rather than a single color temperature or

illuminance value as recommended by WELL, that an ideal circadian lighting system
would change gradually throughout the day
© ASCE
TH
HE IMPACT
T OF LIGH
HTING ON PRODUCT
P TIVITY
Lighting g can effecct productiv vity by makking visual tasks easieer or more
diffficult, by cau
using distracction, and byy manipulatinng the comffort levels off occupants.
Borrisuit, Linhaart, Scartezziini, & Munch h (2014) fouund that subjjects felt sleeepier earlier
in the
t afternoon under elecctric light raather than ddaylight. On a health noote, subjects
alsoo reported loower well-beeing under electric
e lightt. WELL staandards sugggest the use
of controllable
c g within offiice spaces annd the researrch of Juslenn, Wouters,
task lighting
& Tenner
T (20077) suggest th
hat such straategies can inncrease prodductivity by up to 4.5%.
It is
i thought that the sattisfaction of control oover their im mmediate environment
conntributes to this
t productiivity increasee. However,, some of thee research suuggests that
commfort and productivity might actuaally be at oodds. While the most ccomfortable
lighhting conditiions are bettween 401 anda 500lx, thhe optimum m illuminancee for visual
taskks was foun nd to be over 900lx. (Gou, Lau, & Ye, 20014) Mentallly fatigued
inddividuals exp posed to lighht of around 1000lx havee decreased sleepiness aand reaction
tim
me on simple tasks, whilee increasing self-control.. However, oon more com mplex tasks,
the high illumiinance had an adverse effect. (Karrin C. H. J.. Smolders & de Kort,
20114) This ressearch suggeests that a balance
b betwween comfoort and prodductivity be
useed, or periodds of intense illumination n to increasee productivitty with interrvals of low
illu
umination to serve as restt periods.
TH
HE IMPACT
T OF LIGH
HTING ON SATISFAC
S CTION
The liteerature indicaates the link between ligghting and saatisfaction ussed specific
ligh
hting strateg gies or technnologies. A post occupaancy evaluattion (POE) oof the New
Yorrk Times Building
B reveealed occup pants satisfaaction with tthe automatted shading
devvices (Figuree 5) and the automated dimming
d ligghts (Figure 6). Clear, Innkarojrit, &
Leee (2006) stud died occupan nt satisfactio
on with electtrochromic w windows andd found that
Subbjects preferrred variable transmittancce and used venetian bliinds less, butt used more
elecctric lighting
g than fixed transmittancce.
Fig
gure 5. Satisfaction with
h Automated Window Shhades in NY Times Buildding
© ASCE
Figure 6. Satisfactiion with Dim

mmable Lighhting in NY Times Buildding
COUSTIC COMFORT
AC C
This reesearch attempts to qu uantify the occupants’ productivitty benefits,

resuulted from changes
c in acoustic
a quallities, througgh reviewingg the existinng literature
andd comparison n of standarrds. Although often overrlooked in liiterature andd standards,
aco
oustics play a major rolee in occupan nt productivitty. Good acooustics can aallow small
gro
oups to easilly talk without disturbin ng others frrom their wwork, bad accoustics can
keeep work from m getting done
d becausee of constannt distractionn. Acousticss is often a
sub
btle art, so fin
nding studiees on it is diffficult, and m
many of
The staandards thatt specificallly have secctions talkinng about accoustics are
LEED and Weell Building Standard. The T literaturre that talkss specificallyy about the
corrrelation betwween acoustic quality and a perform mance is limmited as welll. Fourteen
artiicles have beeen revieweed in this ressearch. Exceept one that has been puublished on
19997, the articles have been n published between
b yeaars 2006 andd 2014. Beloow there is a
chaart that show
ws the focus ofo the articlees.
Tab ble 1. Maxiimum compo osite sound transmission

t n class ratinggs for adjacent spaces
Ad djacency com mbinations STTCC
Sttandard officce Standard offfice 455
Exxecutive offiice Executive office
o 500
Coonference rooom Conference room 500
Offfice, conferrence room Hallway, staairway 500
Tab
ble 2. Reverrberation tim
me requiremeents
T60 (sec), at 5000 Hz, 1000
Ro
oom type Applicattion Hz, aand 2000 Hzz
Meeting
g or banquet room < 0.88
Offfice buildin
ng Executiv
ve or privatee office < 0.66
Conferen
nce room < 0.66
Teleconfference room
m < 0.66
Open-plan office without sound < 0.88
g
masking
Open-plan office with sound maasking 0.8
© ASCE
Amplified speech < 1.0

Extensive phone use and speech < 0.6
communication
Large-capacity space with
speech amplification < 1.5
In 1972, the EPA presented the Noise Control Act, which defined noise
sources and how to control, maintain, and police noise levels in public areas. While
this doesn’t directly correlate to indoor office standards, it set up a way of thinking
about noise and noise control.
LEED v4 for Interior Design and Construction identifies points to acoustic
performance. Its intent is to provide workspaces that promote occupants’ well-being,
productivity, and communications through effective acoustic design. All occupied
spaces must meet the following design requirements, as applicable, for HVAC
background noise, sound isolation, reverberation time, and sound reinforcement and
masking.
Achieve maximum background noise levels from heating, ventilating, and air
conditioning (HVAC) systems per 2011 ASHRAE Handbook, HVAC Applications,
Chapter 48, Table 1; AHRI Standard 885-2008, Table 15; or a local equivalent.
The WELL Building Standard does not include acoustics. It addresses,
however, to include several standards dealing with acoustics, within the comfort tier.
Most of the research however, deals with partition height and room layouts, leaving
very little quantitative data. However, by using WELL as a base case, more articles
can be found by searching for the specific metrics that WELL utilizes. These can then
be used to evaluate WELL and LEED.
CONCLUSIONS
There has been significant work over the years on the development of case
studies for various aspects of indoor environmental quality (i.e. air, visual, thermal
and acoustic quality.). For the most part, such studies have been stove-piped in the
sense that they focus deeply on only one component. Excellent monitoring tools have
been developed and results have been reported in the literature.
OBSERVATIONS
This work is somewhat unique it its emphasis has been on the combined
effects of the different components of IEQ. It is also unique in its attempt to create
metrics focused on business performance. Given that energy tends are such a small
percentage of operating costs, and one that is effectively transparent to most owners
in commercial office environments, such metrics can become useful in re-focusing
questions about energy performance to improved value and business outcomes.
• Most of the research on occupant satisfaction performance and productivity
has focused on assessing one factor of comfort. Temperature, for example, is
seen as the primary indicator of occupant thermal comfort. But, occupants are
exposed to multiple factors simultaneously, such as air movement, humidity
© ASCE
and raddiant sourcess of heat. Th he weightedd values of tthe interrelattionships of

these issues are imp portant to thee matter of ddefinitively.
• In the Well-Buildiing Standarrd, there arre suggestioons about hheating and
cooling systems. However,
H thee impact off these diffeerent system ms including
personaal comfort co ontrol on prroductivity, satisfaction,, health as w
well as cost
have noot been docum mented.
• It is diffficult to acctually comppare the datta through lliterature revview alone.
Becausee of how thee research wasw conducted and presennted. This suuggests that
increasiing the focuss on better asssessment toools, methodds of data colllection and
interpreetation of meeasured impacts in a connsistent and relative maanner that is
requiredd.
w of the con
Review ntemporary literature suuggests thatt there is a large gap
betw
ween IEQ comfort categ gories from which cohessive conclussion regardinng occupant
welll-being may
y be extrapollated from im
mperial physsiological stuudies.
Ven ntilative Lum

minous Acouustic
Theermal Com
mfort
Com mfort Com
mfort
Commfort
Fig
gure 7. Perccentage of literature
l ad
ddressing W
Well Standardds for Com
mfort, Mind,
Nouurishment an
nd Fitness arreas.
NE
EXT STEPS
S
This liteerature searcch supports the observattion that a mmore cohesivve approach
reg
garding behaavioral influeences and thheir role in aarticulating m
more accuratte occupant
sch
hedules in simmulation toools is neededd. These queestions have been deriveed from this
view and are guiding a seecond phase of this IEQ project.
rev
• Can a means
m of asssessing hierrarchical vallue to the m multiple influences that
influencce comfort be b developeed by introdducing occuppants to thee impact of
these vaariables? Workshops,
W poolicy assessmment and caase study willl be guided
by curreent social sccience metho ods for deterrmining occcupant underrstanding of
the influ
uence upon comfort
c percception.
• Can a cross referenttial matrix ofo thermal syystems compponents be developed to
address up-fit alternnatives in rettrofit and rem
modeling proojects?
• A stand dardized forrmat can bee developedd for assesssing occupaant comfort
sufficienncy? If so, can
c it result in a change in the way ooccupant schhedules and
systemss parameters are formulaated in simullation tools?
© ASCE
Fig
gure 8. Meaasured occup pant energy demand proofile Simulaated vs. actuual – issues
Pollicy and behaavioral undeerstanding off plug load im
mpacts.
AC
CKNOWLEDGEMENT
TS
This research was made possiible throughh funding bby the NSF I/UCRC –
Cennter for Susttainably Inteegrated Build
dings and Siites and our industry parrtners Wells
Farrgo Bank, Baank of Amerrica and Ingeersoll Rand C Corporation.
This reesearch was compiled by b Researchh Assistancce from the School of
Arcchitecture, College
C of Arts
A and Architecture;
A UNC Charrlotte. Ourr thanks to
Niccholas Bradfford, Kelsey Lane, Sheid da Hosseinzaadeh, and NNazanin Moddaresahmadi
for their consid
derable effortt.
BIB
BLIOGRAP
PHIC COM
MPENDIUM
M OF LITER
RATURE R
REVIEW
Thermal Com mfort. Alan Hedge,

H Daniiel E. Gaygeen, 2010; Baaron, R. A., Rea, M. S.,
& Daniels,
D S. G. 1992; B.. W. Olesen n, J. Toftum and J. Kolaarik; Choi JJH, Aziz A,
Lofftness V. 20
010; Gail S. Brager; Geo o Clausen, D David P. Wyyon, 2008; HHui Zhang,
Edw
ward Arens,, DongEun Kim,K Elena Buchbergerr, Fred Baum man, Charliee Huizenga,
200
09; Ioan Sarbbu, Calin Seebarchievici, 2013; J. Koolarik, B. W. Olesen, J. T
Toftum and
Lorrenzo Mattaarolo, 2007; June J. Piilcher, Eric Nadler andd Caroline Buch,2002;
Kassper L.Jen nsen, Jørn nToftum, PeterFriis-H Hansen, 22009; Komalanathan
Vim
malanathan, Thangavelu u Ramesh Babu,
B 2014; Kwok Wai Tham, Hennry Cahyadi
Willem, 2010; K. W. Tham, T 20044; LEENA M. REINIIKAINEN, 2001; Lia
Chaatzidiakou, Dejan
D Mum movic, Julie Dockrell
D , 22014; Li Huaang, Yingxiin Zhu, Qin
Ouyyang, Bin Cao,
C 2012; Li L Lan, Paweel Wargocki , Zhiwei Liaan,2012; Li Lan, Pawel
Waargocki, Zhiwwei Lian, 20011; Li Lan,, Pawel Warrgocki, Zhiw wei Lian, 2014; Li Lan,
Zhiiwei Lian, Li
L Pan, Qian n Ye, 2009; Li Lan, Zhhiwei Lian, Li Pan, 20110; Li Lan,
Zhiiwei Lian, 2009;
2 Lisje Schellen,
S Wouter van M Marken Lichttenbelt, Marrtin de Wit,
Maarcel Loomaans, Arjan Frijns
F and Jørn
J Toftum m, 2008; MMatthew Frannchetti and
Ghorban Komaaki, 2012; M. M J. Mendeell, A. Mirerr, 2009; Naooe Nishiharaa, Shin-ichi
Tan
nabe, Masao oki Haneda,, Masanori Ueki, Akihhiro Kawam mura and Koouei Obata,
© ASCE
2007; Olli Seppänen, William J Fisk, David Faulkner, 2003; Olli A Seppänen,
William J Fisk, 2006; Pawel Wargocki, David P. Wyon ,2013; Pawel Wargocki,
David P. Wyon,2007; R. Kosonen, F. Tan,2004; SINGH, A., SYAL, M.,
KORKMAZ, S. and GRADY, S., 2011; Tanabe, Haned & Nishipana, 2007; Tanabe
S., Kobayashi, K., Kiyota, O., Nishihara, N.,and Haneda, M. 2009; T. Kershaw, D.
Lash, 2013; Valeria De Giuli, Roberto Zecchin, Livio Corain, Luigi Salmaso, 2014;
Weilin Cui, Guoguang Cao, Jung Ho Park, Qin Ouyang, Yingxin Zhu, 2013; W. J.
Fisk, D. Black, G. Brunner,2011; Zs. Bakó-Biró, D.J. Clements-Croome, N.
Kochhara, H.B. Awbia, M.J. Williams, 2012.
Ventilative Comfort. ASHRAE. 2010a. “Thermal Environmental Conditions for

Human Occupancy”, ASHRAE Standard 55-2010, Atlanta, Georgia, American
Society of Heating, Refrigerating and Air- Conditioning Engineers; Ackerly, K.,
Brager, Gail, and Arens, Ed (2012). "Data Collection Methods for Assessing
Adaptive Comfort in Mixed-Mode Buildings and Personal Comfort Systems." Indoor
Environmental Quality 51.; Arens, E. A., A. G. Blyholder and G. E. Schiller (1984).
"Predicting thermal comfort of people in naturally ventilated buildings." HVAC
System 14.; "ASHRAE Standard 62.1-2010." Chair, SSPC 62.1ASHRAE Standard
62.1 Overview (n.d.) n. pag. ASHRAE NEO. Web.
http//www.neo.ne.gov/home_const/iecc/pdf/ASHRAE621.pdf; Brager, G., Barker,
Lindsay (2009). "Occupant Satisfaction in Mixed-mode Buildings." Building
Research and Information 37(4) 18.; Brager, G., Alspach, Peter, and Nall, Daniel H.
(2011). "Natural vs. Mechanical Ventilation and Cooling." RSES Journal HVAC
System 6.;Brager, G. a. B., Sam (2010). "Comfort Standards and Variation in
Exceedance for Mixed-Mode Buildings." Envelope Systems 18.;Brager, G. S. (2001).
"Climate, comfort, & natural ventilation a new adaptive comfort standard for
ASHRAE standard 55." Indoor Environmental Quality 19.; Brager, G. S., E. Ring and
K. Powell (2000). "Mixed-mode ventilation Hvac meets Mother Nature." HVAC
System 7.;Bordass, W. and A. Leaman. 1993. “User and Occupant Control in
Buildings.” Proceedings of the International Conference on Building Design,
Technology and Occupant Well-Being in Temperate Climates. Brussels Feb 17-19.
Borgeson, S. & G. Brager. 2011. “Comfort standards and variations in exceedance for
mixed-mode buildings.” Building Research and Information 39(2) 118-
133.Clements-Croome DJ, Roberts BM. Airconditioning and ventilation of buildings.
2nd ed. Oxford (United Kingdom) Pergamon Press; 1981.Clements-Croome DJ.
Creating the productive workplace. 2nd edition. London Routledge; 2005.Clements-
Croome Dj, P. (2008). "Work Performance, Productivity, and Indoor air." SJWEH
Suppl 4 10.de Dear, R. J. and Brager, G. S. 2002. “Thermal comfort in naturally
ventilated buildings revisions to ASHRAE Standard 55”, Energy and Buildings, 34(6)
549-561Fang, L., Clausen, G, and Fanger, P.O. 1998, “Impact of temperature and
humidity on the perception of indoor air quality during immediate and longer whole-
body exposure”. Indoor Air, Vol. 8, 276-284. Federspiel, C. C., G. Liu, M. Lahiff, D.
Faulkner, D. L. Dibartolomeo, W. J. Fisk, P. N. Price and D. P. Sullivan (2002).
"Worker performance and ventilation Analyses of individual data for call-center
workers." 8. Fisk, William J. "Health And Productivity Gains From Better Indoor
Environments And Their Relationship With Building Energy Efficiency." Annu. Rev.
© ASCE
Energy. Environ. Annual Review of Energy and the Environment 25.1 (2000) 537-66.
Web. Fritsch, R., A. Kohler, M. Nygard-Ferguson, and J.L. Scartezzini. 1990. “A
stochastic model of user behaviour regarding ventilation.” Building and Environment
25 (2) 173-181. Heiselberg, P.K. (2002) Principles of Hybrid Ventilation, Hybrid
Ventilation Centre, Aalborg University, Aalborg. Heinzerling, D., Webster, Tom,
Schiavon, Stefano, Anwar, George, and Dickerhoff, Darryl (2013). "A Prototype
Toolkit for Evaluating Indoor Environmental Quality in Commercial Buildings."
HVAC System 29. Henze, G.P., Felsmann, C., Kalz, D.E. and Herkel, S. (2008)
Primary energy and comfort performance of ventilation assisted thermo-active
building systems in continental climates. Energy and Buildings, 40(2), 99–111.
Honnekeri, A., Pigman Margaret C, Zhang, Hui, Arens, Edward, Fountain, Marc,
Zahi, Yongchao, and Dutton, Spencer (2014). "Use of Adaptive and Thermal
Comfort in a Naturally Ventilated Office." Indoor Environmental Quality 9. Hui
Zhang, D. K., Edward Arens, Elena Buchberger, Fred Bauman, and a. C. Huizenga
(2008). "COMFORT, PERCEIVED AIR QUALITY, AND WORK
PERFORMANCE IN ALOW-POWER TASK-AMBIENT CONDITIONING
SYSTEM.” "Increased Ventilation." LEED User. N.p., n.d. Web.
<http//www.leeduser.com/credit/nc-2009/IEQc2. Indoor Air Quality Scientific
Findings Resource Bank.” Indoor Air Quality Scientific Findings Resource Bank
Health and Economic Impacts of Building Ventilation Ventilation Rates and Office
Work Performance. N.p., n.d. Web. 07 May 2015.
<http//www.iaqscience.lbl.gov/vent-office.html>. Jonsson, A. (2013). "THE VALUE
OF VENTILATION FROM THE WEBER-FECHNER LAW." Productivity and
Economics 8. Kats, G., L. Alevantis, A. Berman, E. Mills, and J. Perlman. 2003. The
Costs and Financial Benefits of Green Buildings. Report to California’s Sustainable
Building Task Force, pp.54-55. Ncube M, Riffat S. Developing an indoor
environment quality tool for assessment of mechanically ventilated office buildings in
the UK – A preliminary study. Building and Environment 2012;5326–33. Olli
Seppänen, W. J. F., QH Lei (2006). "VENTILATION AND PERFORMANCE IN
OFFICE WORK." Indoor Air Journal 18 15. Pasut, W. a. D. C., Michele (2012).
"Evaluation of Various CFD Modelling Strategies in Predicting Airflow and
Temperature in a Naturally Ventilated Double Skin Façade." Applied Thermal
Engineering 37 27. Pawel Wargocki, D. P. W., and P Ole Fanger (2000).
"Productivity is Affected by the Air Quality in Offices." International Centre for
Indoor Environment and Energy 6. Schiavon, S., Melijove, Arsen, and Chandra,
Sekhar (2010). "Energy Savings Strategies with Personalized Ventilation in Tropics."
Energy and Buildings 42(5) 16. Schiavon, s., Bauman, Fred, Tully Brad, and
Rimmer, Julian (2011). "Air Change Effectivness in Laboratory Test of Combined
Chilled Ceiling and Displacement Ventilation." Indoor Air 7. Seppänen OA, Fisk WJ,
Mendell MJ. (1999) Association of ventilation rates and CO2- concentrations with
health and other responses in commercial and institutional buildings. Indoor Air J, 9,
252-274. Seppänen, O. and W. Fisk. 2001. “Association of ventilation system type
with sick building symptoms in office workers”, Proceedings Indoor Air, pp. 98-112.
Seppanen, O. (2008). "Scientific basis for design of ventilation for health,
productivity and good energy efficiency." Indoor Air Congress 8. Spindler, H. and
Norford, L. (2009a) Naturally ventilated and mixed-mode buildings – Part I Thermal
© ASCE
modeling. Building and Environment, 44(4), 736–749. Vivian Loftness, F., Volker
Hartkopf, Ph.D.2, Beran Gurtekin, Ph.D. (2002). "Linking Energy to Health and
Productivity in the Built Environment." Advanced Building Systems Integration
Consortium 12. Wargocki P, Seppanen O, Anderson J, Boerstra A, Clements-
Croome DJ, et al. Indoor climate and productivity in offices guide book 6. Brussels
Federation of European Heating and Air-Conditioning Associations (REHVA); 2006.
WJ Fisk, P. P., D Faulkner, D Sullivan, D Dibartolomeo, C Federspiel, G Liu, and M
Lahiff (2002). "Productivity and Ventilation Rate Analyses of Timeseries data for a
Group of Call-Canter Workers." 6. Li Y, Leung GM, Tang JW, Yang X, Chao CYH,
Lin JZ, et al. Role of ventilation in airborne transmission of infectious agents in the
built environment—a multidisciplinary systematic review. Indoor Air. 2007;17(1)2–
18.
Luminous Comfort. Aries, 2005; Badia, Myers, Boecker, Culpepper, & Harsh,
1991; Baron, Rea, & Daniels, 1992; S. Begemann, G. Van den Beld, & A. Tenner,
1997; S. H. A. Begemann, G. J. van den Beld, & A. D. Tenner, 1997; Berman,
Jewett, Fein, Saika, & Ashford, 1990; Borisuit, Linhart, Scartezzini, & Munch, 2014;
P. Boyce, Hunter, & Howlett, 2003; P. R. Boyce, 2010; P. R. Boyce, Beckstead,
Eklund, Strobel, & Rea, 1997; P. R. Boyce & Cuttle, 1990; Peter R Boyce et al.,
2006; Cajochen, 2007; Cajochen et al., 2005; Cajochen, Zeitzer, Czeisler, & Dijk,
2000; Chellappa et al., 2011; Clear, Inkarojrit, & Lee, 2006; Daurat et al., 1993; De
Carli, De Giuli, & Zecchin, 2008; Deguchi & Sato, 1992; Dijk & Archer, 2009;
Dikel, Burns, Veitch, Mancini, & Newsham, 2013; "<eScholarship UC item
3km3d2sn.pdf>," ; Escuyer & Fontoynont, 2001; Farley & Veitch, 2001; Ferlazzo et
al., 2014; M. G. Figureueiro, Bierman, Plitnick, & Rea, 2009; Mariana G Figureueiro,
Bullough, Bierman, Fay, & Rea, 2007; Fotios & Cheal, 2010; Golden et al., 2005;
Gou, Lau, & Ye, 2014; Heerwagen & Heerwagen, 1986; Hoffmann et al., 2008;
Hubalek, Brink, & Schierz, 2010; Hwang & Jeong Tai, 2010; Ju, Chen, & Lin, 2012;
Juslen, Wouters, & Tenner, 2007; K. Kaida et al., 2006; K. Kaida, Takahashi, &
Otsuka, 2007; Kosuke Kaida, Takeda, & Tsuzuki, 2012; Kantermann et al., 2012;
Kasper, Rogers, Madden, Joseph-Vanderpool, & Rosenthal, 1990; Katzev, 1992;
Lehrl et al., 2007; Leichtfried et al., 2010; Leichtfried et al., 2015; Lewy, Wehr,
Goodwin, Newsome, & Markey, 1980; Linhart & Scartezzini, 2011; Liu, Chiang, &
Lin, 2010; Logadóttir, Christoffersen, & Fotios, 2011; Martinez-Nicolas, Ortiz-
Tudela, Madrid, & Rol, 2011; McLntyre, Norman, Burrows, & Armstrong, 1989;
Menzies & Wherrett, 2005; Mills, Tomkins, & Schlangen, 2007; Moore, Carter, &
Slater, 2003; Munch, Linhart, Borisuit, Jaeggi, & Scartezzini, 2012; Myers & Badia,
1993; Newsham et al., 2009; "Office Worker Response to Fluorescent Lamps of
Different CCT and Lumen Output," ; Partonen & Lonnqvist, 2000; Revell, Arendt,
Fogg, & Skene, 2006; Roche, Dewey, & Littlefair, 2000; Rodriquez & Pattini, 2012;
Rosenthal, Rotter, Jacobsen, & Skwerer, 1987; Rüger, Gordijn, Beersma, de Vries, &
Daan, 2006; Sahin & Figureueiro, 2013; Sahin, Wood, Plitnick, & Figureueiro, 2014;
Scheer, van Doornen, & Buijs, 1999; Sivaji, Shopian, Nor, Chuan, & Bahri, 2013; K.
C. Smolders, de Kort, & Cluitmans, 2012; Karin C. H. J. Smolders & de Kort, 2014;
K. C. H. J. Smolders, de Kort, & van den Berg, 2013; Stephenson, Schroder,
Bertschy, & Bourgin, 2012; Tabuchi, Matsushima, & Nakamura, 1995; Trinder,
© ASCE
Armstrong, O'Brien, Luke, & Martin, 1996; Tuaycharoen & Tregenza, 2007; van
Bommel, 2006; Vandewalle et al., 2006; Vandewalle, Gais, et al., 2007; Vandewalle,
Maquet, & Dijk, 2009; Vandewalle, Schmidt, et al., 2007; JA Veitch, Geerts, Charles,
Newsham, & Marquardt, 2005; JA Veitch & Newsham, 2000; J. Veitch, Newsham,
Boyce, & Jones, 2008; Jennifer A. Veitch, 2001; Jennifer A Veitch, 2001; J. A.
Veitch & Gifford, 1996; Jennifer A Veitch, Hine, & Gifford, 1993; J. A. Veitch &
McColl, 2001; J. A. Veitch, Stokkermans, & Newsham, 2011; Viola, James,
Schlangen, & Dijk, 2008; Wang & Boubekri, 2010; Webb, 2006; Wei et al., 2014;
Wells, 1965; Yildirim, Akalin-Baskaya, & Celebi, 2007; Zeitzer, Ruby, Fisicaro, &
Heller, 2011.
Acoustic Comfort. Salter, C. M. P., & Lawrence, T. R. (2012). Acoustical

performance measurement protocols for commercial buildings. CBE Summary
Report. Salter, C., Powell, K., Begault, D., & Alvarado, R. (2003). Case studies of a
method for predicting speech privacy in the contemporary workplace. Center for the
Built Environment. Kim, J., & de Dear, R. (2013). Workspace satisfaction The
privacy-communication trade-off in open-plan offices. Journal of Environmental
Psychology, 36, 18-26. Goins, J., Chun, C., & Zhang, H. (2012). User perspectives on
outdoor noise in buildings with operable windows. Bauman, F., Carter, G.,
Baughman, A., & Arens, E. A. (1997). A field study of PEM (Personal
Environmental Module) performance in Bank of America's San Francisco office
buildings. Center for the Built Environment. Heinzerling, D., Schiavon, S., Webster,
T., & Arens, E. (2013). Indoor environmental quality assessment models A literature
review and a proposed weighting and classification scheme. Building and
Environment, 70, 210-222. Lee, Y. S., & Guerin, D. A. (2009). Indoor environmental
quality related to occupant satisfaction and performance in LEED-certified
buildings. Indoor and Built Environment, 18(4), 293-300. Kim, J., & de Dear, R.
(2012). Nonlinear relationships between individual IEQ factors and overall
workspace satisfaction. Building and Environment, 49, 33-40. Abbaszadeh, S.,
Zagreus, L., Lehrer, D., & Huizenga, C. (2006). Occupant satisfaction with indoor
environmental quality in green buildings. Center for the Built Environment. Huang,
L., Zhu, Y., Ouyang, Q., & Cao, B. (2012). A study on the effects of thermal,
luminous, and acoustic environments on indoor environmental comfort in
offices. Building and environment, 49, 304-309. Lee, Y. S. (2010). Office layout
affecting privacy, interaction, and acoustic quality in LEED-certified
buildings. Building and Environment, 45(7), 1594-1600. Collinge, W. O., Landis, A.
E., Jones, A. K., Schaefer, L. A., & Bilec, M. M. (2014). Productivity metrics in
dynamic LCA for whole buildings Using a post-occupancy evaluation of energy and
indoor environmental quality trade-offs. Building and Environment, 82, 339-348.
Satisfaction and self-estimated performance in relation to indoor environmental
parameters and building features Liang, H. H., Chen, C. P., Hwang, R. L., Shih, W.
M., Lo, S. C., & Liao, H. Y. (2014). Satisfaction of occupants toward indoor
environment quality of certified green office buildings in Taiwan. Building and
Environment, 72, 232-242.
© ASCE
Direct Measurement of Occupants’ Skin Temperature and Human Thermal

Comfort Sensation for Building Comfort Control
Pooya Sharifani1; Suraj Talele1; Junghyun Mun1; and Yong Tao1*

1
University of North Texas, Department of Mechanical and Energy Engineering,
3940 North Elm St., Room F101A, Denton, TX 76207-7102. E-mail:
yong.tao@unt.edu
Abstract
Providing conditioned and fully controlled indoor environment is one of goals of

building design. But on the other hand, smart building controllers are providing the
required cooling and heating load depending on occupants’ real time feelings. This
study focuses on evaluating meaningful and steady state relationship between human
thermal perception and body physiological behavior and express comfort feeling
perception as a measurable and sensible variable in general occupants through their
ambient. For this purpose different body part reaction in the same ambient situation
has been studied and two body parts which have the biggest linear temperature
changes to ambient condition, has been selected to evaluate this research hypothesis.
For these tests, wrist and palm temperatures on different people have been measured
accurately with thermal camera to follow the temperature trend on various comfort
levels which has been caused by different ambient condition. It has been founded that
each person has his own unique temperature on wrist and palm when he feels comfort
in the ambient he is in. In the other words each person’s body temperature at comfort
level is a precise indicator for comfort feeling of his own. So in the future by having
this unique comfort temperature for each person, controlling the indoor condition
with the occupants’ body temperature to approach them to their unique hand comfort
temperature would be a rational choice to bring convenience while energy has been
saved more.
INTRODUCTION
Thermal comfort has become a new talking point when people consider energy
efficiency in the residential/commercial buildings. The most complicated and
dynamic part in energy modelling is thermal comfort due to dynamic nature of human
actions and unpredictable nature of human behavior. But, it seems that modelling
human thermal comfort may be possible by different techniques as demonstrated by
Cao et al. (Cao, 2011) by comparing Thermal sensation vote (TSV) to Predicted
mean vote (PMV) and using ASHRAE index of 7 levels to categorize thermal
comfort. Thermal comfort can be studied as function of air temperature (Cui, 2013)
and dynamic modelling of thermal comfort using IMEM model is demonstrated by
Katavoutas et al. (Katavoutas, 2015). Various other modelling approaches have been
studied and described in literature survey below. Here, the intention is to find
© ASCE
correlation between human palm temperature and comfortable temperature of an

individual and it has been tried to validate if palm temperature can be used as an
independent reliable parameter to gauge human thermal comfort. The experiments
were carried out in summer season in a 1200 ft2 residential test lab called Zero
Energy Lab (ZOE) at Discovery Park campus of University of North Texas. Five sets
of experiments were performed and 25 experiments were conducted in total with 5
participants. Currently, all participants were male and ages ranged from 27 yrs to 35
yrs. All participants were students and healthy males. The experiments were
conducted with Flir ThermaCam B2 infrared camera which offers non-invasive
means of monitoring and diagnosing the high resolution imagery. The paper presents
a detailed literature survey followed by methodology followed for analysis and then
data analysis is presented which is supported by appropriate discussion and finally
conclusions are demonstrated.
LITERATURE REVIEW
Cao et al studied thermal comfort and thermal adaptability in a university setting

in Beijing, China. The study was conducted on summer and winter seasons and the
thermal comfort was gauged based on ASHRAE Index having 7 levels from -3 to +3.
204 students participated in the study, 82 in summer and 124 in winter (Cao, 2011).
The Chinese territory was divided into North and South for this study. The operative
temperature was used to determine thermal comfort and thermal adaptability study.
The authors found that in summer clothing insulation is low and that is why there is
little adjustment feasibility. The standardized thermal sensation vote (TSV) and
Predicted mean vote (PMV) models were created in order to develop relationship
between operative temperature and thermal sensation (Cao, 2011). The authors
conclude that in winter, the thermal sensation was always bigger than PMV
prediction. They also conclude that in winter outdoor temperature influenced thermal
adaptability of the person. The authors found that keeping indoor temperature high in
winter leads to wastage of energy and this in turn makes the incumbent person
uncomfortable. The authors also conclude that native regional climate affects thermal
adaptability of people living there (Cao, 2011). Cui et al studied human thermal
comfort as a function of air temperature. They performed experiments based on
environment with five different temperatures of 22 ºC, 24 ºC, 26 ºC, 29 ºC and 32 ºC
(Cui, 2013). The experiments were performed in a climate chamber of 5X 4X 3 m. 36
participants were used for experiments and were divided into Group A and B. 18
males and 18 females participated. Group A was exposed to all five temperatures
while Group B was exposed to 26 ºC environment. LSI BABUC data logger was used
for recording air temperature and air velocity while TSV was used for thermal
comfort. Work load was measured by NASA Task Load Index while statistical
analysis was performed using General Linear Model Repeated Measures procedure
(Cui, 2013). The authors conclude that learning effect is affected by temperature and
that changing temperature and thermal discomfort reduce learning speed. Warm
discomfort enclosure has bad effect on performance compared to colder discomfort
© ASCE
(Cui, 2013) and motivation for high performance increases with high thermal comfort
(Cui, 2013). Dynamic modelling of thermal comfort has been studied by Katavoutas
et al. They used two node IMEM model for dynamic modelling of thermal transients.
Temporal pattern of body temperatures, skin wittedness and water loss were the
parameters investigated (Katavoutas, 2015). The authors concluded that thermal
adaptation and response to the same neutral thermal environment is different for
individuals based on their activities outdoors (Katavoutas, 2015). The skin
temperature was lower when individual sits under shade (Scenario A) compared to
temperature when individual is under direct solar radiation and sitting (Scenario B)
and also skin temperature was lower than (Scenario C) where individual was walking
(Katavoutas, 2015).
Kolarik et al. simulated thermo active building system (TABS) and compared the
results with performance of all-air VAV ventilation system. The TABS systems are
essentially the hydronic heating/cooling systems also known as radiant
heating/cooling systems. The building modules were considered with heavy and light
construction with two different orientations (east-west and north-south) (Kolarik,
2011). The authors found that TABS saved 50% more primary energy than VAV
system for hot-humid climate, thermal sensation expectations did not change more
than 1% for 60%-70% 0f working hours for TABS system (Kolarik, 2011). Change
of building orientation did not have significant impact on primary energy use while
90% of occupants were thermally satisfied for using TABS system in moderate and
hot-dry climate (Kolarik, 2011). Li et al. performed an experiment in a naturally
ventilated room in Chongqing, China. 300 students were part of the experiment and
the authors found that sensory nerve conduction velocity (SCV) was fine measure of
human response assessment for thermal comfort (Li, 2010). SCV can produce some
adaptive changes in human body to deal with indoor temperature changes due to
adaptive feedback regulation (Li, 2010). Luo et al. conducted field studies in Beijing
and Shanghai, China to find out influence of personal control on thermal comfort.
139 apartments were used for experiments. Occupants with personal control had
neutral temperature lower than occupants with no personal control by 2.6 ºC (Luo,
2014). A review of field studies has been presented by Mishra et al. They conclude
that conditioned areas have restricted thermal comfort zones and that similar thermal
environments are perceived differently by occupants based on their activities
outdoors, ease and effectiveness of the systems (Mishra, 2013). Cultural habits like
sitting on cool floors and opening windows which help thermal comfort in naturally
ventilated buildings are limited to cultures and regions associated (Mishra, 2013).
Moustris et al. developed and applied Artificial Neural Network (ANN). The model
were based on bioclimatic data calculated from Discomfort Index and Cooling Power
Index and microclimatic data like air temperature, humidity etc. The data was
considered for 2001-2005 (Moustris, 2010). The ANN models can successfully
predict thresholds for human discomfort (Moustris, 2010). Conditions of heatstroke in
humans and days with high thermal discomfort can be predicted through ANN
models (Moustris, 2010). Authors suggest to focus on developing operational
forecasting system for urban areas to predict severe human discomfort (Moustris,
2010). Ng et al. studied urban thermal comfort in Hong Kong China. The authors
© ASCE
found that wind speed of 0.9-1.3 m/s and physiological equivalent temperature (PET)
of 28 ºC is enough for Hong Kong public to experience thermal comfort (Ng, 2012).
Authors suggest proper urban planning, building coverage and proper layout (Ng,
2012). Authors also suggest limiting solar radiations in the streets by providing
canopies, covered streets, walkaways, recesses, etc. (Ng, 2012).
Rabi et al. performed exergy analysis by developing a thermodynamic model for

thermoregulatory heat transfer for a human sitting with light clothed body (Rabi,
2012). The PMV and PPD are not sufficient for predicting human comfort and that
developing and irreversibility based discomfort index is necessity for predicting
human thermal comfort (Rabi, 2012). The authors found that clothing influenced heat
transfer mechanisms convective or radiative affect irreversibility rate (Rabi, 2012).
Irreversibility losses can be reduced when heat is transferred convectively more than
radiative manner (Rabi, 2012). Tseliou et al. investigated correlation between PET,
temperature humidity index (THI) and wind chill index (K) and evaluated them by
comparing to Actual thermal sensation votes (ASV). The study was performed in 7
European cities. The authors reported deviation of different correlations when
compared to ASV thermal comfort class (Tseliou, 2010). They found strong relation
between climatic mean temperature with the indices and concluded projected
deviation of thermal comfort. The authors suggested developing a new index which
closely associated ASV based sub-set and its value correctly projects comfort
(Tseliou, 2010). Vanos et al. model human exercising in outdoors using four-segment
skin temperature with COMFA (COMfort FormulA) outdoor energy balance model
(Vanos, 2012). The authors conclude that COMFA accurately predicts the mean skin
temperature with deviation of 1.5 ºC (Vanos, 2012). Spearmans rank correlation (rs)
was used by authors to evaluate relationship between ATS and mean skin temperature
and its value was 0.507 and 0.517 (Vanos, 2012). The authors conclude that
individuals were more tolerant to thermal discomfort when accustomed to outdoor
conditions with physical exercise (Vanos, 2012).
EXPERIMENT AND METHODOLOGY
Due to the uncertainty to model general human behavior, new approach to

correlate building condition with occupant’s sensation is necessary. So it is required
to find out a way to report human sensation as a measurable parameter. Most of the
authors have focused on temperature of different human body parts such as [face,
nose, ear, neck, chest, foot] because they have the biggest range of temperature
difference in different building indoor conditions. The only physical property which
has significant influence on human comfort is temperature and it has rigorously
investigated in literature. In general the factors which influence on human thermal
sensation are: air temperature, relative humidity, mean radiant temperature and air
speed. Men are more sensitive to temperature while women mostly are more sensitive
to relative humidity. Different factors which cause variation of the comfort levels are
age, gender, ethnicity, level of clothing, moment of the day and moment of the year
that the experiment is taking place.
© ASCE
The present study were performed on five participants and all of them were
students. All participants were male and the experiment was conducted in summer.
Total 25 experiments were performed in five sets. Participants wore short sleeve t-
shirt and light jean pants which has clothing insulation value of 0.34 Clo together. All
the experiments were performed in Zero Energy Lab (ZOE) at University of North
Texas on roughly consecutive days around 2:30 PM over the June 2015. ZOE is a
unique residential test lab thermally controlled by radiant floor system and water-to-
water heat pump (WWHP) systems with forced air. Room conditions during
experiments were checked dynamically through building management system
TACVista. Ambient conditions were maintained at 25.5 °C and the relative humidity
was variable between 73%-77% which effectively changed the dew point temperature
between 21 °C – 22 °C. Fig.1 shows the Flir ThermaCam B2 infra-red camera used
for measurements of palm and wrist temperatures during experiments. It is possible to
set emissivity and reflective temperature of human skin in the ThermaCam. Human skin
emissivity is 0.98 and reflective temperature is 33.33 °C due to the closest temperature to
the average participant’s hand temperature. Reflective temperature setting is provided in
the thermal camera to compensate for the temperature reflected from the objects
surrounding the subject of the experiment.
The experiment protocol was implemented in following manner: first each

participant was asked about his comfort level in the normal lab condition based on
ASHRAE 7-level index from -3 to +3. Where -3 corresponds to very cold and +3
corresponds to very hot and 0 being neutral. Due to heavy activities or any routine
activities, if the participants were not in neutral comfort zone then they were requested to
rest for at least 15 minutes. During this time, the participants were allowed to use fans,
coolers or any means in order to make them comfortable and bring their comfort feeling
to level 0. The experimental measurements were taken two times, once when the
participant was not using any fan and was his feeling was just depend on the lab ambient
temperature which usually was uncomfortable (i.e. level 2). The other measurement was
performed when the participant was in neutral comfort zone (i.e. level 0) which has been
provided by using fan to cool them down until feel neutral. The experiments were
performed by targeting the thermal camera at two precise spots on participants’ body.
Middle of the right hand palm and center of the wrist, right where watch band is locked
as shown in Fig.2. The physical position of the participant during the experiment is
shown in Fig.3. This process was repeated 25 times for five participants.
© ASCE
Figu
ure 1. Flir Therm
maCAM B2 Fig
gure 2. Skin temp
mperature measuurement Figgure 3. Position of the subjects
RE
ESULTS AN
ND DISCUS
SSION
Results aree presented for correlatiion betweenn human therrmal comforrt sensation
and
d palm and wrist
w skin tem
mperatures for
f 5 particippants in Figuure 4 to Figurre 8.
© ASCE
Figure 4. Skin temperature Figure 5. Skin temperature measurement

measurement of 1st participant of 2nd participant
Figure 6. Skin temperature Figure 7. Skin temperature

measurement of 3rd participant measurement of 4th participant
Figure 8. Skin temperature measurement of 5th participant
© ASCE
The x-axis denotes comfort level while y-axis shows temperature of the body
part in °C. Because the experiment was done in summer season, ASHRAE index of 0
to +3 for comfort level, is plotted in the graphs by neglecting the -3 index as it is
relevant during cold climate.
The experiment set out to evaluate the possible correlation between human
thermal comfort feeling and selected body parts temperature and also if wrist is a
better spot than palm for capturing the skin temperature. The diagrams illustrate that
the wrist temperature is always less than palm temperature. It also can be seen that
they are convergent together when participants tend to feel warm. Also in the same
feeling levels, the temperature difference between the coldest test and warmest test is
not big on the both spots. Different experiment sessions give similar results and they
could represent precise correlation between comfort feeling level and the two spots
surface temperature on hand. In addition, wider temperature difference on the wrist
which can also be explained by grater slope trend line, makes the wrist relatively
better spot for measuring the whole hand’s part skin surface temperature. Error
analyzing for the temperature variation at 0 level comfort sensation at the biggest
case was 0.7 °C for the palm and 0.8 °C for the wrist. Regarding to the necessity of
repeating this experiment, for future research it is planned to perform the same
experiment in winter season to check the consistency in body physiological behavior
with these results. Although the ambiguity of age, gender and race influence on these
results still exist in the supposed hypothesis, but for next series of experiments,
number of participants will increase and different age groups and genders would be
considered for analysis.
CONCLUSION
Wrist and palm were the potential candidates for evaluating the possible
correlation between human comfort perception and body physiological behavior, due
to their feasibility in making simple apparatus to measure that and also their
sensitivity and accountability to ambient condition changes. It is assumed that in each
thermal comfort level which reported by the participants, each person has his own
individual skin temperature. By over 25 experiment series, the results are in
acceptable concordance with the elementary hypothesis and they show that alongside
rise and falls in skin temperature on different comfort sensation, each person has his
own unique skin temperature on each comfort level scale. Diversely, from this
illustration it can be branched out that by learning each person’s unique wrist skin
index temperature at his neutral or best comfort level, the room condition can be
provide cooling or heating load based on mean time wrist skin temperature deviation
from the comfort index temperature. So there would not be excessive energy injection
to the indoor environment anymore. Also for more than one person in this system, the
average of their comfort index temperature can be login to the system as the set point
temperature for average occupants’ skin temperature and then the central unit HVAC
system provide required load based on the real time difference between these two
values.
© ASCE
REFERENCES
Cao, B. (2011). “Field study of human thermal comfort and thermal adaptability
during the summer and winter in Beijing” Energy and Buildings, 43,1051-1056.
Cui, W. (2013).“Influence of indoor air temperature on human thermal comfort,

motivation and performance” Building and Environment, 68,114-122.
Katavoutas, G.(2015).“Dynamic modeling of human thermal comfort after the

transition from an indoor to an outdoor hot environment” International Journal of
Biometerology, 59, 205-216.
Kolarik, J. (2011). “Simulation of energy use , human thermal comfort and office
work performance in buildings with moderately drifting operative temperatures”
Energy and Buildings, 43, 2988-2997.
Li, B. (2010). “Physiological expression of human thermal comfort to indoor

operative temperature in the non-HVAC environment” Indoor and Built
Environment, 19(2), 221-229.
Luo, M. (2014). “Can personal control influence human thermal comfort? A field
study in residential buildings in China in winter” Energy and Buildings, 72, 411-418.
Mishra, A.K. (2013).“Field studies on human thermal comfort- An overview”

Building and Environment, 64, 94-106.
Moustris, K. P. (2010). “Artificial neural network models as a useful tool to forecast

human thermal comfort using microclimatic and bioclimatic data in the great Athens
area (Greece)” Journal of Environmental Science and Health, 45(4), 447-453.
Ng, E. (2012). “Urban human thermal comfort in hot and humid Hong Kong” Energy
and Buildings, 55, 51-65.
Rabi, J.A. (2012). “Human thermal comfort: an irreversibility-based approach

emulating empirical clothed-body correlations and the conceptual energy balance
equation” Journal of the Brazilian Society of Mechanical Science and Engineering,
34(4), 450-458.
Tseliou, A. (2010). “An evaluation of three biometeorological indices for human

thermal comfort in urban outdoor areas under real climatic conditions” Building and
Environment, 45, 1346-1352.
Vanos, J.K. (2012).“Thermal comfort modelling of body temperature and

psychological variations of a human exercising in an outdoor environment”
International Journal of Biometeorology, 56, 21-32.
© ASCE
Incorporation of Future Building Operating Conditions into the Modeling of

Building–Microclimate Interaction: A Feasibility Approach
Kelly Kalvelage1; Ulrike Passe2,4; Caroline Krejci3; and Michael C. Dorneich1,3

1
Human Computer Interaction; 2Department of Architecture; and 3Department of
Industrial and Manufacturing Systems Engineering, Iowa State University, Ames,
Iowa. 4E-mail: upasse@iastate.edu
Abstract
This paper presents a novel modeling methodology that integrates the near
building environmental conditions (or microclimate), whole-building design, and
occupant behavior. Accurate predictions of the future building operating conditions
lead to designs that serve the building’s purpose – to support occupants’ tasks. This
study bridges the gap between human factors and architecture to include physical,
cognitive, and organizational systems into building information modeling using
future typical meteorological year climate data, canyon air temperature microclimate
model, and a whole-building energy simulation to investigate the impact of future
microclimate conditions on a “typical” single-occupant office. Additionally, to
capture the effects of building occupant decision-making and adaptive behaviors, an
agent-based model is proposed. Model inputs are task-based, which aim to produce a
more robust model to investigate a variety of human-building control interactions to
ensure high building performance and occupant comfort and satisfaction.
INTRODUCTION
Building operating conditions are the product of the dynamic and complex
interactions between the climate, building, and occupants. Modeling these
interactions can significantly improve building performance and occupant comfort.
However, failing to properly represent these interactions leads to buildings that do not
operate as intended and do not satisfy occupants (Andrews, Yi, Krogmann, Senick, &
Wener, 2014). Hence, it is important to accurately account for the potential factors
influencing building operating conditions so that buildings can be better designed to
serve the purpose and needs of building occupants. Efforts to improve accuracy of
commercial building performance predictions has led to increased attention on
modeling building occupant behavior and microclimate – two main drivers of
building energy use (Bonte, Thellier, & Lartigue, 2014; Bouyer, Inard, & Musy,
2011).
Building modeling overview. The commercial building design phase frequently uses
building simulations to examine the interactions between climate, building, and
occupants. It is common practice to rely on stand-alone building configurations in
© ASCE
building simulations. However, it is well known that the urban microclimate strongly
affects building operation (Erell & Williamson, 2007; Santamouris et al., 2001). The
urban microclimate is the local atmospheric zone modified by energy exchanges with
local built environment context. Buildings in urban context experience higher
ambient temperatures due to different wind flow patterns caused by building
placement and form, use of materials that effectively store short-wave radiation, and
increased solar radiation because of high-albedo materials on neighboring surfaces.
Microclimate models simulate these interactions to adapt climate weather data from a
stand-alone meteorological station to realistic site-specific air temperatures.
The building interior includes building systems, furniture, equipment, and
occupants. These are typically represented in building simulation as “oversimplified,
predetermined inputs” that are unrepresentative of actual building systems and
occupancy (Lee & Malkawi, 2014). Inputs are reduced to averages that are then
applied to the whole building, an approach that does not represent the dynamic
environment and behaviors observed in a real workspace. Equipment use, light use,
occupant presence, and occupancy count, to name a few, are reduced to fixed
schedules and represented as fixed metabolic heat generators based on historical data
(Mahdavi, 2001), or as a discrete “stimulus-behavior” relationship (Reinhart, 2004).
Further, furniture that affects airflow and radiant heat exchanges are treated as a fixed
mass that is evenly distributed throughout the spaces.
Future conditions are challenging to predict. It is even more challenging to
predict operating conditions for a building that do not yet exist – for example, a
building designed in a new development where no other structures yet exist. On the
other hand, advancements toward more accurate predictions has led to research
identifying probable operating circumstances (Burian, McPherson, Brown, Streit, &
Turin, 2003). For instance, urban planning describes the context in which individual
buildings are constructed, such as building form, building height-to-width ratios, and
street orientation (Elmualim, Valle, & Kwawu, 2012). Climate-based design
strategies can relate to envelope design, façade materials, and climate mitigation
strategies, such as reflective glazed facades and high albedo materials, to radiate,
conduct, and convect the climate to produce desired result of building performance
(EPA, 2013; Schuetter, DeBaillie, & Ahl, 2014). Further, tasks occupants perform in
a building inform various aspects of building interiors: the furniture, equipment,
interior environment conditions required to carry out day-to-day business operations,
and building systems needed to achieve those operating conditions (Leach, Lobato,
Hirsch, Pless, & Torcellini, 2010). Additionally, tasks can provide expected occupant
activity and clothing levels.
To account for uncertainties in occupant behavior, agent-based modeling
(ABM) has gained increasing attention in building simulation (Lee & Malkawi,
2014). ABM is a computer simulation of agents that interact with their environment
through defined rules, and have the ability to learn and adapt in response to changes
in the environment and other agents. Existing ABMs have utilized either an occupant
satisfaction driver of behavior (Andrews et al., 2014) or an environmental affordance
driver (Lee & Malkawi, 2014). While both have the capability to address variability
in human behavior, these models lack generalizable results because there is no
© ASCE
relaationship beetween thesee behaviors and the truue driver off occupant bbehavior in
commmercial buiildings – thee tasks occup pants perform m.
The obj
bjective of th
his paper is to
t present ann integrated modeling m methodology
thatt takes a task-based approach
a to define parrameters forr building design and
occcupant behaavior in a commercial
c building wwhile accounnting for m microclimate
inflluences on occupant comfort
c connditions andd energy uuse in builddings. This
appproach bridges the gap between
b humman factors aand architectture to includde physical,
coggnitive, and organizationnal systems into
i buildingg informatioon modeling. By basing
model inputs on o tasks rath her than occcupants, the goal is to pproduce a m more robust
model that ev valuates futuure building operatingg conditionss in terms of energy
connsumption anda occupannt comfort. This paperr, first, pressents a dataa collection
metthodology, based in humanh facto
ors methodss, to collect and organnize future
buiilding operaating condittions. Colleccted data w will then bee used to inform the
miccroclimate, building,
b andd occupant components
c iin building ssimulation.
DA
ATA COLLE
ECTION METHODOL
M LOGY
Dettailed informmation is reequired abo out microcli mate, buildding, and occcupants to
acccurately reprresent futuree operating conditions.
c F
Figure 1 outtlines a hum
man factors-
bassed methodology to captu ure data for a Midwest ccommercial ooffice buildiing.
Figuree 1. Research
h methodology used to develop a m
model of thee office
domain.
d
First, an
a ethnographic study y of the office dom main providdes a full
undderstanding of the occup pants and their tasks by documentinng people’s behavior in
the context off their own space. Seco ond, the chaaracterizatioon of buildinng systems
idenntifies and represents
r bu
uilding systeems in a way
ay that highliights both fuunction and
cap
pability. Thirrd, analysis of human behavior
b reppresents the link betweeen occupant
beh
havior, build ding perform
mance, and occupant
o peerformance. The results from each
metthod will be b analyzed d separatelyy, and integgrated into abstractionn hierarchy
dessigned to reflect the coonstraints (i.e. social, phhysical) of the work environment
usinng an Ecolo ogical Interface Design n (EID) appproach, as ooutlined by Burns and
Hajjdukiewicz (2004).
(
Ethhnographic study of the office dom main. Severaal techniques will be useed to gather
task
k-based info
ormation requ
uired to defiine a typical office buildding:
• Review w of comfort standard ds will identify the bbaseline connditions for
occupan nt comfort in an office environm ment. Standaards includee ASHRAE
(2013) Standard 555: thermal en nvironmentaal conditionss for humann occupancy
© ASCE
and EN (2007) Standard 15251: indoor environmental input parameters (air

quality, thermal environment, lighting, and acoustics) for design and
assessment of energy performance of buildings.
• Thermal comfort surveys will assess overall comfort level within occupant
workspace in relation to actual building thermal conditions.
• Contextual task analysis questionnaire twill collect a more granular level of
occupant preferences in terms of their tasks performed.
Responses from each method will be organized into an ontology that defines the
types, properties and interrelationships between the components. Tasks will be
characterized by criticality, frequency, and priority. The physical structure (furniture,
equipment, and spatial characteristics) will be organized by task requirements. The
physical environment (thermal, visual, acoustical, and air quality) associated with
each task will be compared to the actual measurements. For example, if one occupant
reported thermal conditions as uncomfortable during a task, and the task was
performed near an exterior wall, the recorded temperature of the office at that
location would be considered too low, and a warmer condition would be desired.
Finally, personal factors (layers of clothing and preferences) will be considered when
comparing reported conditions to measured conditions.
Characterization of building systems. A preliminary systematic review of existing

case study and literature research was conducted to identify building systems used in
office buildings. Field notes will be taken during the ethnographic study to document
the building systems found in the buildings of the questionnaires. Preliminary
building system components were identified and organized into categories based on
characteristics of capability and function. Component categories were reviewed to
include only the systems most common or important to building operation, based on
four criteria: function, safety, impact on the occupant, and building efficiency.
Analysis of human behavior. A literature review of human behavior research will be

performed to identify influencing factors and effects on occupant comfort,
performance, and satisfaction in the office domain. The theoretical framework
established by Feige, Wallbaum, Janser, and Windlinger (2013) to connect building
criteria, comfort parameters, and performance parameters. Preliminary behavior
research was analyzed for influential factors related to tasks, and linked to the
outcome effects. Factors include attitudes such as experience, opportunity, and
expectations; social norms such as organizational rules and social status; and
environmental influences such as disruptions and annoyances. Outcome effects
include measures of productivity, satisfaction, and building energy usage.
Model of the office domain. To construct the model of the office domain, four steps
were followed. First, the system boundary and scale were defined to include anything
users might want to control or have information about, and exclude anything
redesigned in future work, such as system controls and sensors. Second, an
abstraction hierarchy was created using a part-whole decomposition to develop the
© ASCE
oveerall work do
omain structture. Lastly, the abstracttion hierarchhy model wiill be tested
for completeness using scennarios in whhich the occuupants perforrm various ooffice tasks.
Tassk definitioon. Preliminary results from the coontextual tassk analysis reveal four
inittial common n tasks for examinatio on: 1) electtronic docum ment preparration (e.g.
wriiting or typin
ng a documeent on a commputer), 2) noon-electronicc document preparation
(e.g
g. reading or writing ussing pen and d paper), 3) phone call,, and 4) sm mall meeting
(e.g
g. 2-4 occupants gathered in the samme vicinity foor discussionn). These tassks were the
most frequently performed d, critical to
o the occuppant’s job deescription, aand are the
lon
ngest in durration when being perfformed. Furrther, these four tasks all require
diffferent enviro
onmental coonditions. Using
U these ffour tasks, bbuilding moodel aspects
andd occupant behavior are determined.
BU
UILDING SIIMULATIO
ON METHO
ODOLOGY
Y
Thiis section wiill present th

he applicatio
on of the moodel of the ooffice domaiin and tasks
to each
e compon nent in the building simu ulation proceess outlined in Figure 2.
Figure 2. Simulation
n process.
Microclimate. The micrroclimate conditions aare the prooduct of tthe canyon

morphology an nd regional weather daata. The urbban canyon is defined by a street
(cannyon floor)) bordered by buildin ngs on bothh sides (caanyon wallss). Canyon
morphology deefines the sitee and buildinng characterristics. Charaacteristics innclude street
orieentation, cannyon dimen nsions, streeet and buildding materiaal properties, moisture
d anthropogenic heat. The regional weather dataa for this woork will use
avaailability, and
the future typiccal meteorolo
ogical year (FTMY)
( datta for Chicaggo, Illinois, constructed
usinng the metthod describ bed by Pattton (2013)). This metthod evaluaates typical
metteorological year (TMY Y3) data reco orded at O’HHare Internaational Airpoort for total
sky
y cover, dry--bulb temperature, dew--point tempeerature, relattive humiditty, absolute
hummidity, presssure, and wiind speed foor the periodd 2041 to 20070. A regioonal climate
model was pairred with a gllobal climatee model to ggenerate a m moderate clim mate change
scenario for Chicago.
C Th
he canyon morphologyy and regional weathher will be
com
mbined using g the Canyon
n Air Tempeerature modeel, created bby Erell and W Williamson
(20
006), to geneerate the microclimate conditions
c fo
for the Near West Side community
area in Chicago o.
Building. The Departmentt of Energy’s new consttruction com mmercial meddium office
ogy (Deru et al., 2011) seerves as the larger buildiing structuree in which a
buiilding typolo
© ASCE
sing
gle-occupannt office is deefined. A sin
ngle-occupannt office is ccentrally loccated east to
wesst and alongg the north wall
w on the second
s levell because it w
was identified as being
the most affectted by the suurrounding microclimate
m e, based on results from
m a previous
stud
dy (Kalvelag ge, Passe, & Dorneich, 2015).
2 The m model of thee office dommain defines
the single-occu
upant office ini terms of size,
s shape, layout, furniture, and eqquipment in
ord
der to perforrm the four tasks, as shown in Figgure 3. Addditionally, thhe building
systems for thee single-occuupant office include
i overrhead suppliied heating aand cooling,
a th
hermostat wiith a limited
d range of co ontrol, a mannual window w roll shade, and a door
lock
kset.
Figuree 3. Single-o
occupant offfice schemattic.
Occupant. Usiing NetLogo o, an agent-bbased modell (ABM) off the office ddomain will
be developed. Occupants will be reepresented bby autonom mous agentss that have
inteelligence (i.ee., internal lo
ogic), as welll as the abillity to make complex deecisions and
enggage in co omplex inteeractions wiith other aagents and objects w within their
envvironment. The office domain ABM A will simulate occcupant behhavior and
inteeractions thaat occur betw ween buildin ng occupantss and their ooffice environnment as a
fun
nction of tasks performeed (e.g., the agent is perrforming a hhigh activityy level task
andd adjusts its clothing lev vel according gly). The AABM will alsso account ffor dynamic
occcupant adapttations, enab bling agents tot “learn” ovver time andd respond prooactively to
their environmeent (e.g., thee agent loweers the windoow shade beefore leavingg at the end
of the
t day to prevent early y morning su unlight fromm overheatingg the office tomorrow).
Furrther, occupaant actions do not neceessarily havee a guaranteeed effect onn the office
envvironment, an nd the ABM M will allow this
t uncertaiinty to be caaptured.
ABM accommodat
a es the repreesentation off heterogeneeous occupaants via the
devvelopment of multiple different
d ageent types, orr personas. E Each occupaant persona
willl be characterized by a set of uniqu ue, innate peersonal attribbutes and dyynamic state
varriables derivved from th he ethnograp phic study aand the hum man behavioor analysis.
Atttributes incllude objectives, preferrences, expectations, ssatisfaction thresholds,
assigned tasks, prior knowlledge, attitud des, motivatiions, and exxpertise. Statte variables,
which represen nt an agent’’s state at anya point inn time throuughout the simulation,
desscribe an ageent’s currentt task behaviiors (e.g., thee agent is onn the phone)), as well as
outtcomes that depend
d uponn personal attributes
a andd the state oof its environnment (e.g.,
the agent is com mfortable, giiven that the office is at a certain tem mperature).
© ASCE
The attributes and current state of each individual agent will impact its ability
to perform required tasks, and may affect other occupant environmental
characteristics within the building, thereby influencing other agents’ states and
behaviors. This dynamic and adaptive process of interactions between the building
and its occupants will lead to system-wide performance over time, which will be
measured in terms of energy consumption and occupant productivity and comfort.
Whole-building energy simulation (BES). EnergyPlus (DOE, 2013) will be used to

conduct whole building energy simulations. Using microclimate conditions, building
typology, and site information, the simulations evaluate thermal and visual comfort
and building energy demand. Thermal comfort will be evaluated in terms of predicted
mean vote (PMV), predicted percentage of dissatisfied (PPD), and mean radiant
temperature (MRT). The PMV is the average thermal sensation response from a
larger number of subjects and is the most recognized thermal comfort model. PPD
correlates to PMV. However, because comfort is subjective, there will be a
distribution of satisfaction among a large group of people. This satisfaction or
dissatisfaction is represented by PPD. MRT is related to the amount of radiant heat
transferred from a surface and on the receiving material’s ability to absorb or emit
heat. Visual comfort will be evaluated for glare index and daylight autonomy. Glare
index is the degree of discomfort glare. Daylight autonomy is the percentage of
annual daytime hours that, at a given point in a space, is above a specified
illumination level. Energy demand will be evaluated in terms of heating, cooling, and
total energy use, peak heating and cooling demand, lighting energy, and equipment
energy.
DISCUSSION AND CONCLUSION
The current methodology to evaluate energy efficient building operations (with preset
schedules, set points, and occupation patterns) does not integrate occupancy behavior
nor specific climate features of the near building environment. The prediction, thus,
lacks depth and precision, buildings do not perform as predicted, and occupants
operate buildings differently than anticipated by the designing architecture team
(Andrews et al., 2014). With dominant societal goals to develop high performance
and energy efficient buildings, it has become necessary to develop more refined
models for building performance prediction. This paper presented a new
methodology that incorporates future building operating conditions, as well as
predictions of occupant behavior through an agent-based model, into the modeling of
building-microclimate interactions. The purpose is to improve energy and comfort
design predictions for buildings in urban context and in a changing climate. The
research goal of this project is to provide the inadequate modeling capacity that
enables more refined predictions based on occupant interactions with the building and
occupant task comfort levels. The resulting methodology can be leveraged to further
define model inputs that are more robust to long-term changes and more
generalizable across different building types.
© ASCE
REFERENCES
Andrews, C., Yi, D., Krogmann, U., Senick, J., & Wener, R. (2014). Designing
buildings for real occupants: An agent-based approach. IEEE Transactions on
Systems, Man, and Cybernetics - Part A: Systems and Humans, tba. doi:
10.1109/TSMCA.2011.2116116
ASHRAE. (2013). ASHRAE Standard 55: Thermal environmental conditions for
human occupancy. In R. a. A.-C. E. American Society of Heating, Inc. (Ed.):
ASHRAE.
Bonte, M., Thellier, F., & Lartigue, B. (2014). Impact of occupant's actions on energy
building performance and thermal sensation. Energy and Buildings, 76(0),
219-227. doi: http://dx.doi.org/10.1016/j.enbuild.2014.02.068
Bouyer, J., Inard, C., & Musy, M. (2011). Microclimatic coupling as a solution to
improve building energy simulation in an urban context. Energy and
Buildings, 43(7), 1549-1559.
Burian, S., McPherson, T., Brown, M., Streit, G., & Turin, H. J. (2003). Urban
environmental modeling and assessment using detailed urban databases. Earth
Science in the City: A Reader, 56, 303-333. doi: 10.1029/056SP 12
Burns, C., & Hajdukiewicz, J. (2004). Ecological interface design: CRC Press.
Deru, M., Field, K., Studer, D., Benne, K., Griffith, B., Torcellini, P., . . . Crawley, D.
(2011). U.S. Department of Energy commercial reference building models of
the national building stock. Golden, Colorado: National Renewable Energy
Laboratory.
DOE. (2013). EnergyPlus (Version 8.2.7) [Computer software]: U.S. Department of
Energy (DOE). Retrieved from www.energyplus.gov
Elmualim, A., Valle, R., & Kwawu, W. (2012). Discerning policy and drivers for
sustainable facilities management practice. International Journal of
Sustainable Built Environment, 1(1), 16-25. doi:
http://dx.doi.org/10.1016/j.ijsbe.2012.03.001
EN. (2007). EN 15251: Indoor environmental input parameters for design and
assessment of energy performance of buildings addressing indoor air quality,
thermal environment, lighting and acoustics. In E. C. f. S. (EN) (Ed.).
Brussels: EN.
EPA. (2013). Reducing urban heat islands: Compendium of strategies. Retrieved
from http://www.epa.gov/heatisland/resources/compendium.htm.
Erell, E., & Williamson, T. (2006). Simulating air temperature in an urban street
canyon in all weather conditions using measured data at a reference
meteorological station. International Journal of Climatology, 26, 1671-1694.
doi: 10.1002/joc.1328
Erell, E., & Williamson, T. (2007). Intra-urban difference in canopy layer air
temperature at a mid-latitude city. International Journal of Climatology, 27,
1243-1255. doi: 10.1002/joc.1469
Feige, A., Wallbaum, H., Janser, M., & Windlinger, L. (2013). Impact of sustainable
office buildings on occupant's comfort and productivity. Journal of Corporate
Real Estate, 15(1), 7-34. doi: http://dx.doi.org/10.1108/JCRE-01-2013-0004
Kalvelage, K., Passe, U., & Dorneich, M. (2015). Investigating the impact of
reflective facades on the microclimate. Building and Environment.
© ASCE
Leach, M., Lobato, C., Hirsch, A., Pless, S., & Torcellini, P. (2010). Technical
support document strategies for 50% energy savings in large office buildings
(pp. 141). Golden, CO.: National Renewable Energy Laboratory.
Lee, Y. S., & Malkawi, A. M. (2014). Simulating multiple occupant behaviors in
buildings: An agent-based modeling approach. Energy and Buildings, 69(0),
407-416. doi: http://dx.doi.org/10.1016/j.enbuild.2013.11.020
Mahdavi, A. (2001). Simulation-based control of building systems operation.
Building and Environment, 36(6), 789-796. doi:
http://dx.doi.org/10.1016/S0360-1323(00)00065-2
Patton, S. L. (2013). Development of a future typical meteorlogical year with
application to building energy use. (Master of Science Thesis), Iowa State
University, Ames, Iowa. (Paper 13635)
Reinhart, C. F. (2004). Lightswitch-2002: a model for manual and automated control
of electric lighting and blinds. Solar Energy, 77(1), 15-28. doi:
http://dx.doi.org/10.1016/j.solener.2004.04.003
Santamouris, M., Papanikolaou, N., Livada, I., Koronakis, I., Georgakis, C., Argiriou,
A., & Assimakopoulos, D. N. (2001). On the impact of urban climate on the
energy consumption of buildings. Solar Energy, 70(3), 201-216. doi:
http://dx.doi.org/10.1016/S0038-092X(00)00095-5
Schuetter, S., DeBaillie, L., & Ahl, D. (2014). Future Climate Impacts On Building
Design. ASHRAE Journal, 56(9), 36-44.
© ASCE
Measuring the Effectiveness of an Immersive Virtual Environment for the

Modeling and Prediction of Occupant Behavior
Sanaz Saeidi1; Tracy Rizzuto2; Yimin Zhu3; and Robert Kooima4

1
Department of Engineering Science, Louisiana State University, 246 Old Forestry
Building, Baton Rouge, LA 70803. E-mail: ssaeid1@lsu.edu
2
School of Human Resource Education and Workforce Development, Louisiana State
University, 282 Coates Hall, Baton Rouge, LA 70803. E-mail: trizzut@lsu.edu
3
Department of Engineering Science, Louisiana State University, 214-A Old Forestry
Building, Baton Rouge, LA 708030. E-mail: yiminzhu@lsu.edu
4
School of Electrical Engineering and Computer Science, Louisiana State University,
2016 Digital Media Center, Baton Rouge, LA 70803. E-mail: kooima@csc.lsu.edu
Abstract
The main purpose of this paper is to determine a procedure or methodology

for validating the effectiveness of immersive virtual environment (IVE) applications
to studies of occupant behavior. In recent decades, IVE experiments have become
increasingly popular in various fields of study such as healthcare, education, science
and engineering. They have been effectively utilized to replicate field experiences
that are costly or risky in research contexts. Furthermore, this paper proposes to use
IVE in initial building designs to identify the most significant predictors of users’
lighting-use behavior. The authors are interested in exploring the extent to which
occupants’ responses in IVE accurately reflect those in-situ. In fact, the ultimate goal
of this study is to enhance IVE research designs for measuring occupants’ energy
behaviors and integrate the results with the existing energy simulation models and
reduce the uncertainties in estimates.
Keywords: Building energy performance; Occupant lighting use behavior;
Immersive virtual environment.
INTRODUCTION
Human Energy-related Behavior. During the past decades, many studies have
focused on energy efficiency in built environments and the demand for more
sustainable buildings. Even though the deterministic features of the building energy
simulators appear to be quite useful (Haldi and Robinson, 2008), the randomness of
occupant-related factors and the complexity of human nature make it rather hard to
accurately predict occupant behavior. Clevenger and Haymaker (2006) argued that
accuracy and precision of existing energy simulation models are still uncertain since
predicting the actual performance of buildings is highly dependent upon users’
© ASCE
behavior. Also, “initial results show that predicted energy consumption changes by
more than 150% using all high or all low values for what experts believe reasonably
represents occupant behavior”. Thus, understanding the way that occupants interact
with buildings is a contributing factor to determining energy saving potentials.
Immersive Virtual Environment (IVE). Recent advancements in computer display

technology have opened up an opportunity to place individuals within an Immersive
Virtual Environment (IVE) and let them physically move and interact with virtual
objects. “Its attraction lies in the tendency for individuals to react in virtual reality as
they would in the real-life situation”(Sanchez-Vives and Slater, 2005). Far beyond
that, IVE experiments have been effectively utilized to replicate field experiences that
are costly or risky in research contexts. IVE integration with health-related research
settings has resulted in successful interventions such as stress management (Serino et
al., 2014), testing delusional beliefs (Freeman et al., 2010), treating paranoid
disorders (Freeman et al., 2008) and acrophobia (Hodges et al., 1994). Furthermore,
IVE has acquired an excellent reputation in educating individuals to operate complex
industrial equipment. Tichon (2007) in his study of high-stress training of train
drivers in a virtual environment, reported simulators’ effectiveness in generating
“reasonable fit with the real world”.
Similarly, in a study by Messner et al. (2006) , the replication of IVE with
field experiments has proved to be able to improve the total building energy
performance due to the simulation of different building environmental systems.
Likewise, Dunston et al. (2011)offer an IVE mock-up tool to review the design of the
space and equipment functionality through added interaction capabilities of the mock-
up practices. Also, in a study of modeling traffic patterns, using virtual reality was
demonstrated as an effective technique to observe the way people show specific
patterns of interaction, and to identify the indicators on how to establish rules in
empirical procedures for decision-making (Tan and De, 2000).
IVE Validation Methods. There have been many studies to date focusing on the
validity of IVE experiments and its measures in order to determine its value as a
measuring tool. The major validity types that should be considered in IVE
experiments are given as below.
Ecological validity. Loomis et al. (1999) state that “IVE technology is a highly
promising tool for the study of basic psychological processes. Its primary advantages
are affording more ecological validity without compromising experimental control
and allowing the decoupling of variables that naturally co-vary”. When it comes to
measuring ecological validity, presence or sense of “being there” becomes an
absolutely inevitable concept to consider. The term presence is generally used to
address the human feelings and perceptions of a virtual environment (Barfield et al.,
1995). The reason behind finding participants’ sense of being there is that the more a
person feels present in a virtual environment, the more his/her response to the stimuli
would match those in the physical environment (Villani et al., 2012).
© ASCE
Along with testing the level of presence in IVE, the face validity of the
experiment has to be confirmed as well, i.e., the experiment material should look like
they are measuring what they intend to measure and be designed as respondent-
friendly as possible (Oppenheim, 2000).
Content validation in IVE is considered as another requirement to determine
the value of IVE as a research tool, which is referred to as the appropriateness of the
simulator in its functionality (Lyons et al., 2013).
Internal validity. According to Loomis et al. (1999)“Internal validity is possibly the

most important strength of IVE studies since the entire VEs can be easily controlled.
Stimulus control and experimental stimulus manipulation is a key feature in
investigating cause and effect relations”.
Construct validity. Construct validity for an IVE test can be assessed by comparing
the applied measures in two different IVE and in-situ settings. If both of the test
settings produce similar measurement and path coefficients, then construct validity
can be supported (Trochim, 2000).
External validity. Supposing that in this experiment there is a causal relationship

between the constructs of the causes and their relevant effects, to provide support for
external validity of the research, researchers should be able to generalize the results
of this study to other persons, places or times. Besides, “General knowledge can only
be produced through application of theory and no findings based on any research
methodology can have external validity in the absence of theory” (Lucas, 2003).
Criterion validity. Finally, IVE effectiveness will be established by demonstrating

comparable levels of prediction in behavioral intentions and occupancy behaviors
between the two settings. Similar predictive capacities between the two settings will
provide evidence of criterion-related validity.
RESEARCH OBJECTIVES
The main focus of this study is to determine a procedure or methodology for

validating the effectiveness of IVE to studies of occupant behavior in buildings. The
authors try to figure out what tests will ensure that data collected from IVE can be
considered valid and comparable to those collected in-situ or lab environments.
Hence, it is necessary to assess the functionality and effectiveness of the Immersive
Virtual Reality (IVR) technology in general as a test tool and confirm the validity of
particular IVE settings as a test environment as well.
© ASCE
Experiment Environment and Procedure. The authors initially tested the

framework of the study in a series of experiments, simulating different lighting
alternatives in-situ and in IVE. The main goal of the experiments was to assess
participants’ visual perception in different conditions and explore their consequent
behaviors with implications of their energy use preference.
Tests in physical setting. The experiment room was located in a typical office
building on the LSU campus; the test environment was equipped with office
furniture, computer and specialized tools for the test and the room was illuminated
with ceiling lights as well as a south-oriented window. The tests were carried out at a
specific time of day with certain default amount of outdoor illumination. Ten students
participated in the experiment. At first they were asked for their background
information and after that the experiment procedures started by the reading task in a
dimly lit condition, which continued by completing a visual perception questionnaire.
It was intended to make a thorough examination of individuals’ visual perception, so
a previously tested questionnaire from the study of Lee et al. (2014)was applied to
carefully evaluate subjects’ visual perception and also to observe their subsequent
lighting use behavior. After the lighting condition of the room was adjusted based on
subjects’ preferred choice, they were asked to perform another reading task and fill
out the visual perception questionnaire one more time. Their feedbacks were recorded
for further evaluations and comparison with an IVE test.
Tests in virtual setting. The second test setting was a 3-D virtual environment that
represented almost all of the visual qualities of the in-situ test environment. The
virtual scene was also able to reflect the equivalent natural outdoor light and the
ceiling lights. The experiment room was simulated with a powerful game design tool,
Unreal Engine 4.6, so that it can give participants the opportunity to manipulate some
physical features of their virtual test environment such as turning the light on/off and
opening/closing the window blind. Participants had no previous experience in IVEs,
so they received training of maneuvering in the virtual scene and the use of the head
mount device and the controller before the experiment starts. They were taught to
explore the IVE demo scenes, adjust the equipment and make changes to the
illumination of the scene in a comparable way to those in-situ. Once they became
completely familiar with the environment, the test started and similar to the in-situ
exercise, subjects performed a reading task in a dimly lit virtual environment. Then,
they answered the post experiment visual perception questionnaire and they were
given options to improve their visual comfort. Subsequently, after changing the
lighting condition of the room, they performed another reading task and filled out the
visual perception questionnaire once again. After the reading tasks were over, the ITC
Sense of Presence Inventory (ITC-SOPI) was utilized to assess some validity aspects
of the experiments (Lessiter et al., 2001) and at the very end of the test, participants
were asked to give their opinions about their behavior related to lighting consumption
through Theory of Planned Behavior (TPB) questionnaires (Ajzen, 2006, Ajzen,
1991) in order that criterion validity of the experiment can be supported.
© ASCE
There are
a four lightting choices in each of tthe physical and virtual test setting,
nammely P1, P2, P3 and P4, which
w respeectively reprresent the exxperiment environment
witth natural and
a artificiall lighting toogether, onlly natural llighting, onlly artificial
ligh
hting and finnally no lig ghting source at all. Th ey are illusttrated in thee following
table (see Tablee 1).
able 1: Experiment Environmentss
Ta
Room Condition
C In-situ photo
p IVE illustration
P1 Blind
d open/
Lig
ght off
P2 Blind
d open/
Ligght on
Blind
d closed/
P3 Lig
ght on
P4 Blind
d closed/
Lig
ght off
EX
XPERIMEN
NTAL RESU
ULTS
Ecoological valiidity of the IVE

I setting has been evvaluated throough self-repports on the
C-SOPI queestionnaire. For the natturalness off the sceness, subjects rreported an
ITC
aveerage of 76%%, ranging from
f 48% to o 92%. The least recordded measuree refers to a
sub
bject who ex xperienced th he highest negative
n imppact of IVE ttools. Almost all of the
parrticipants reported bein ng negativelly impactedd due to thhe imperfecttion of the
IVEE and the eqquipment. Esspecially, if this
t effect w was shown too increase, thhe results of
the visual perceeption were less likely to
o correlate w
with those off the in-situ ttest.
© ASCE
For measuring the sense of presence ITC-SOPI questionnaire pledges to

support it as well. In this exercise the IVE test received high rating on
engagement and spatial presence respectively 76% and 78%. In addition to the
presence, ITC-SOPI questions have been designed with careful considerations to
provide support for the face validity on top (Lessiter et al., 2001). For content
validity, a question regarding IVE reflection was framed and we anticipated that the
majority of test subjects either agreed or strongly agreed that the IVE test was able to
replicate in-situ experience and indeed 70% of the participants responded to this
question as expected.
The threat to internal validity was eliminated by developing completely
comparable scenes, plus by providing adequate and step-by-step explanation about
every single stimulus control and experimental stimulus manipulation in both test
settings.
Construct validity was assessed by comparing the mean performance of the
tested variables, which consist of visual comfort, visual stimulation and mood in two
IVE and in-situ settings. Except the subjects who faced high negative impact of IVE,
both settings produced almost similar measurement and path coefficients, i.e., the
correlations between visual perception measures in P4-in-situ and P4-IVE also Px-in-situ and
Px-IVE (x:1, 2, 3) have been reached enough high to match between their similar
constructs. It seems that lighting stimuli in the IVE were able to produce similar
behavioral responses and the collected data reflected well the construct of the two
experiments.
However, in terms of emotional aspects, IVE showed that it cannot represent
constant and reliable responses. For instance, many subjects felt relax, cozy and
pleasant for the reading task in the dimly lit physical environment, but results show
that no subject could feel the same in the equivalent condition in IVE.
The external validity of the study could not be assessed due to the small
sample size of the experiment. Therefore, further studies are required so that authors
can generalize the findings of the research to and across other groups of people, time
and place.
Lastly, the criterion validity in this study was tested by the demonstration of
similarly predictive TPB measurement model. The TPB measures which consist of
attitudes and beliefs, norms, control beliefs and actual behavioral control have been
used as independent variables in order to predict the behavioral intentions and
subjects’ ultimate behaviors. The TPB measures in this study have shown a great
potential to be the evidence of criterion-related validity as the subjects’ lighting
behavior closely matched their self-report survey. Indeed the applied measures were
likely to predict the behavior of 70% of the participants. More significantly, the
observations revealed that the predictive capacities can possibly get lost when the
subjects experience strong negative effect of the IVE.
© ASCE
With the above observations, the authors propose a set of hypotheses for
further testing, which has been summarized and presented in the following table (see
Table 2).
Table 2: Summary of analysis methods and hypothesis testing
Type of Hypothesis Quality Method for Data Gathering
Validity Assessed
Ecological IVE and the in-situ context Presence Post-experiment questionnaire
can produce a comparable (ITC-SOPI)
level of spatial presence and Face Post-experiment questionnaire
engagement (ITC-SOPI)
Content Post-experiment questionnaire
(Qualitative IVE reflection
inquiry)
Internal IVE and the in-situ context Cause and Effect Observation and experimental
can produce a comparable Relation stimulus manipulation control
cause and effect relation
Construct IVE and the in-situ context Visual Comfort 1) Post-experiment
can produce comparable Visual questionnaire (Visual
measurement and path Stimulation perception)
coefficients Mood 2) Behavioral inquiry
Criterion IVE and the in-situ context Attitude and Post-experiment questionnaire
can result in comparable Beliefs (Theory of planned behavior)
behavioral predictions. Norms
Control Beliefs
Actual
Behavioral
Control
DISCUSSION AND RECOMMENDATION
As discussed before, estimations of building energy simulators range widely

from optimistic assumptions and experts are convinced that the randomness of
occupant-related factors is one of the main reasons for it. Findings of this study
indicated that replication of IVE with field experiments appears to be an extremely
promising method to improve those models. Undeniably, the new adaption will be a
useful and effective adjunct to any building project. Besides, the authors believe that
comparing various design alternatives with specific energy behavioral information
can assist designers and engineers to optimize designs. Nonetheless, this study
remains to be inconclusive due to tis explorative nature and a small sample size. The
authors recommend future research with larger sample size and more advanced
equipment.
REFERENCES
AJZEN, I. 1991. The theory of planned behavior. Organizational behavior and
human decision processes, 50, 179-211.
AJZEN, I. 2006. Constructing a theory of planned behavior questionnaire.
© ASCE
BARFIELD, W., ZELTZER, D., SHERIDAN, T. & SLATER, M. 1995. Presence

and performance within virtual environments. Virtual environments and
advanced interface design, 473-513.
CLEVENGER, C. M. & HAYMAKER, J. The impact of the building occupant on
energy modeling simulations. Joint International Conference on Computing
and Decision Making in Civil and Building Engineering, Montreal, Canada,
2006. Citeseer, 1-10.
DUNSTON, P. S., ARNS, L. L., MCGLOTHLIN, J. D., LASKER, G. C. &
KUSHNER, A. G. 2011. An immersive virtual reality mock-up for design
review of hospital patient rooms. Collaborative Design in Virtual
Environments. Springer.
FREEMAN, D., PUGH, K., ANTLEY, A., SLATER, M., BEBBINGTON, P.,
GITTINS, M., DUNN, G., KUIPERS, E., FOWLER, D. & GARETY, P.
2008. Virtual reality study of paranoid thinking in the general population. The
British Journal of Psychiatry, 192, 258-263.
FREEMAN, D., PUGH, K., VORONTSOVA, N., ANTLEY, A. & SLATER, M.
2010. Testing the continuum of delusional beliefs: an experimental study
using virtual reality. Journal of abnormal psychology, 119, 83.
HALDI, F. & ROBINSON, D. 2008. On the behaviour and adaptation of office
occupants. Building and environment, 43, 2163-2177.
HODGES, L. F., KOOPER, R., MEYER, T. C., DE GRAAFF, J. J. H.,
ROTHBAUM, B. O., OPDYKE, D., WILLIFORD, J. S. & NORTH, M. M.
1994. Presence as the defining factor in a VR application.
LEE, J.-H., MOON, J. W. & KIM, S. 2014. Analysis of Occupants’ Visual
Perception to Refine Indoor Lighting Environment for Office Tasks. Energies,
7, 4116-4139.
LESSITER, J., FREEMAN, J., KEOGH, E. & DAVIDOFF, J. 2001. A cross-media
presence questionnaire: The ITC-Sense of Presence Inventory. Presence, 10,
282-297.
LOOMIS, J. M., BLASCOVICH, J. J. & BEALL, A. C. 1999. Immersive virtual
environment technology as a basic research tool in psychology. Behavior
Research Methods, Instruments, & Computers, 31, 557-564.
LUCAS, J. W. 2003. Theory‐testing, generalization, and the problem of external
validity. Sociological Theory, 21, 236-253.
LYONS, C., GOLDFARB, D., JONES, S. L., BADHIWALA, N., MILES, B., LINK,
R. & DUNKIN, B. J. 2013. Which skills really matter? Proving face, content,
and construct validity for a commercial robotic simulator. Surgical
endoscopy, 27, 2020-2030.
MESSNER, J. I., RILEY, D. R. & MOECK, M. 2006. Virtual facility prototyping for
sustainable project delivery. ITcon.
OPPENHEIM, A. N. 2000. Questionnaire design, interviewing and attitude
measurement, Bloomsbury Publishing.
SANCHEZ-VIVES, M. V. & SLATER, M. 2005. From presence to consciousness
through virtual reality. Nature Reviews Neuroscience, 6, 332-339.
SERINO, S., TRIBERTI, S., VILLANI, D., CIPRESSO, P., GAGGIOLI, A. &
RIVA, G. 2014. Toward a validation of cyber-interventions for stress
disorders based on stress inoculation training: a systematic review. Virtual
Reality, 18, 73-87.
TAN, A. & DE, V. B. 2000. Investigation of Human Behavior in a Virtual
Environment.
TICHON, J. G. 2007. Using presence to improve a virtual training environment.
CyberPsychology & Behavior, 10, 781-788.
TROCHIM, W. M. 2000. The Research Methods Knowledge Base, 2nd Edition.
Atomic Dog Publishing, Cincinnati, OH.
© ASCE
VILLANI, D., REPETTO, C., CIPRESSO, P. & RIVA, G. 2012. May I experience
more presence in doing the same thing in virtual reality than in reality? An
answer from a simulated job interview. Interacting with Computers, 24, 265-
272.
© ASCE
Integrated Project Delivery and Total Building Automation for the Nearly Net-Zero-
Energy Q1 ThyssenKrupp Headquarters
Thomas Spiegelhalter1
1
Co-Director Structures and Environmental Technologies Lab (SET-Lab), College of Architecture + The
Arts, Florida International University, PCA, 11200 SW 8th St., Miami, FL 33199.
E-mail: tspiege@fiu.edu
Abstract
These days, the design and construction industry is challenged with the question “how are
architects, engineers, contractors, building owners using information technology to further
automate their practices, building operation and benchmarking?” The following case study with
interviews will critically evaluate the ‘Integrated Project Delivery’ process of the ‘Q1
ThyssenKrupp Quarter’ in Essen, Germany built from 2009 to 2010, and decommissioned in
2012. The Q1-Energy Concept of the ThyssenKrupp Headquarters’ primary energy consumption
is measured at 139.6 kWh/m²/yr and its ecological footprint has by 27% less CO2 emissions than
similar buildings in this category. One of the major innovative systems are the Solar Facade
controlled system and the Q1-Total Green Building Automation (DESIGO). Siemens specifically
developed a Building Automation System (BAS) called Green Building Monitor™ to supervise
the building consumption data of Q1 continuously for expert analysis. DESIGO includes a
specially developed energy-efficient operating control concept for 1,133 rooms.
Keywords: Sustainability; Net-zero-energy (NET-Zeb); Carbon-neutral; Total building

automation; Human-computer-interface (HCI); Digital mock-up (DMU); European building
energy performance certification.
DESIGN, SIMULATION, FABRICATION AND ASSEMBLY WORKFLOW TOWARDS

TOTAL BUILDING AUTOMATION
The schematic and automated workflow for the integrated project delivery of the Q1
ThyssenKrupp Quarter includes the early design stage, 3-D-simulations, detailing, file-to-
production process, assembly, commissioning, benchmarking, and the total green building
automation. This case study evaluates how the German Thyssen-Krupp headquarters’ real-time
SIEMENS total green building automation system (BAS) performs with intelligent control
feedback loops and learning algorithms for constantly optimized building performance, security,
and user comfort operation. The BAS also includes a wireless environmental management
system to ensure trend analysis and optimizations toward yearly net-zero-energy certifications
based on the mandatory yearly European Building Energy Performance Certification system.
Interviews about the 2-D to 3-D/4-D/5-D processes. The Q1 lead architect J. Steffens of JSWD
Architekten in Cologne, Germany, described that the computationnel shift in his office into 3-D
also paralleled and engaged a conceptual and pragmatic shift which was required to navigate and
© ASCE
manage through increased organizational complexity and required flexibility of the 300,000 m²
gross floor area with 10 different occupancy types and zonings”. According to Steffens, the first
shift in the design process from 2-D AutoCAD / SketchUp to 3-D-Max and Rhino/ Grasshopper
happened when the client demanded a more collaborative and transparent working environment.
The second shift occurred when the complexity of the project demanded realtime 3-D/4-D
simulations with VRfx-compatible formats in OpenGL Performer software to share, reiterate,
synchronize, and visualize quick changes and updates in the parameters with the client and
special contractor direct input. The VRfx open format of the OpenGL Performer was developed
by the German Fraunhofer Gesellschaft (FhG) in collaboration with the Competence Center
Virtual Environments (CCVE) of the Fraunhofer IAO and the Institute for Human Factors and
Technology Management (IAT) of the University of Stuttgart. The spatial 3-D visualization for
direct interaction and immersion into the virtual architecture of Q1 was facilitated in particular
by Caetano da Silva and Prof. Guenther Wenzel from Fraunhofer IAO in Stuttgart for JWSD
Architekten and Chaix & Morel et Associés.
INTEGRATED DESIGN & PROJECT DELIVERY PROCESS
The early ‘integrated design project to fabrication delivery process’ was parallel
supported by the development of the real-time OpenGL Performer analysis and scenario tool to
assist the design phases and also later -after the realization of the headquarter- the real-time
scheduling with the Total Green “Building Automation System (BAS). The integrated BAS
synthesized intelligent control feedback loops and self-learning algorithms for constant
optimized building operation and bench-marking. It was clear at this early level of collaborative
design that the virtual reality tool could be simultaneously a medium, prototype, and rapid
modeling tool to visualize, manipulate data and analyze complex relationships intuitively. The
Fraunhofer Visual Processing of 3-D Content (VRfx) format has similarities with other 3-D
game engines, such as Quest3D® fed by 3-D-data resulting from multiple scenario inputs of
built environment modeling (BEM) tools, to create navigable, immersive, integrated 3-D
architecture models or urban environments using false color. According to Prof. Guenter Wenzel
“these virtual models enhance designers and clients” perception by allowing them to explore
experientially hundreds of potential multidisciplinary design combinations, both qualitatively
and quantitatively, in a practical, collaborative and cost effective way. Participatory processes
with accessible data exchange are essential for getting necessary change and radical design
innovations adopted. Today, designers have the opportunity to create interactive models with
interrelated data that are immersive—including a mix of real and symbolic representation—that
facilitate experiential design and ‘what if’-scenarios’ (Figure 1)
© ASCE
Figure 1: Immersive Open GL Performer Simulation n (Images Coourtesy of G

Guenter Weenzel,
Virtual Environmeents, Institu
ute for Human Factors and Technoology Manaagement, IA AT at
th
he Universitty of Stuttgaart, 2013)
gure 2: Scheematic work

Fig kflow diagram of the in
ntegrated prroject delivvery for Q1
ThyssenKru
T upp Quarteer (Diagram
m courtesy off Thomas SSpiegelhalterr, 2013.)
Design and
a Producction of the intelligen nt multifuncctional Sun n/Daylight Façade Coontrol
System. One of the major
m innov
vative Q1 systems are thhe solar facaade controlleed system annd the
Q1-total green buildding automaation (DESIG GO). The German company Siem mens specifi
fically
developeed DESIGO to operate ass an integratted Building Automationn System (BA AS). This allso so
called Grreen Buildin
ng Monitor™ ™ is created to supervisee in real-tim
me the buildiing operationn and
assemblees continuouusly consum mption data for the exppert and straategy analyssis. For exaample
DESIGO O includes a specially developed
d en
nergy-efficieent operatinng control cooncept for 1,133
rooms. Sensors
S record daylight and detect the
t presencee of people, which allow ws the systeem to
provide optimal
o worrkspace therrmal and lighting condittions based on those reeal-time readdings.
The system displayss accurately to the hourr – how mucch power annd water is consumed iin the
building,, the tempeerature, and greenhousee gas emisssions. BAS monitors uuser humann and
building systems beehaviour, compares, meeasures and visualizes rreal-time progress data with
© ASCE
various performance
p benchmarkss for carbon neutrality oor the nearly zero-fossil eenergy standdards.
Other feaatures includ de integrated control fo or electrical and securitty technologgy tailored tto the
specific and
a dynamicc needs of th he ThyssenK Krupp campuus buildings ((Fig. 3).
The
T Sun and daylight con ntrol system
ms are equippped with suittable glare reemediation w while
still givin
ng the buildiing occupannts sufficientt outviews annd visibilityy require an integrated ddesign
approach h. The archiitects design ned successffully a direcct link to thhe dynamicc and perpettually
evolving patterns of outdoor illu umination an nd direct solaar radiation with a uniqque controlleed but
dynamic and adaptiv ve sensor controlled day ylighting faç ade system. It is beneficial clearly helps
to create a visually sttimulating annd productivve environmment for the bbuilding occuupants whilee also
reducing significantly the total building
b enerrgy costs annd greenhousse gas emisssions (Fig. 4). All
the Q1 office
o m steel profilles and operrable verticaal lamellas. In total therre are
façadees have slim
400,000 stainless steeel slats coveering an areaa of approxim
mately 8,0000 m² to contrrol sun and
Figuree 3: Images ab
bove: Q1 Th
hyssenKrupp p Quarter witth automated d solar and d
daylight conttrol
system, JSWD Arch hitekten and Chaix & Mo orel et Associiés, 2006. (Im
mage: Thomaas Spiegelhallter,
2013.) Images beloow: Q1 lamellla performance angles peerpendicularr to the glasss façade. Imaage
right, bottom:
b The virtual
v anim
mation of the façade
f by Frrener + Reifeer with digitaal 3-D mock-ups
(So
ource: Freneer + Reifer, S
Sept. 2013.)
daylight entering thee building. This

T prototyp pe system rem mains fully functional eeven in high wind
speeds of 70 km/hou ur, which is unusual forr highrise buuildings. Thhis prototypee system rem
mains
© ASCE
fully functional even in high wind speeds of 70 km/hour, which is unusual for highrise buildings.
Most outside shading and daylight systems in similar building types around the world turn off
during the high-speed wind season, and as a result their energy consumption increases
dramatically due higher cooling loads. The solar and daylighting control concept was
schematically designed by JSWD Architekten and Chaix & Morel et Associés and
algorithmically detailed and optimized through multiple simulations and digital mock-ups
(DMUs) by the Fraunhofer Institute for Solar Energy Systems (FhG) in Freiburg, Germany in
collaboration with the special façade contractor Frener+Reifer from South Tyrol in South Tirol,
Italy. The DMUs included a parametric algorithmic dynamic weather data-driven solar radiation
and illumination analysis with multizone simulations to engineer a constant horizontal overhang
for summer sun protection and daylight control. All systems were combined with a vertical set of
twisting fins. The matte underside of the system was further designed by the architects and
engineers and Frener+Reifer after the FhG modeled and optimized various property behaviors
such as reflection, shadow, daylight, glare, and heat protection (Fig. 4). All the shading elements
are sensor controlled through a separate EIBSKNX- Bus system and embedded for data
recording in the SIEMENS BAS with the assistance of 1,600 electric motors to dynamically
perform to the following settings: Closed or parallel to the glass façade Moving with the sun /
variable and perpendicular to the sun Open: the horizontal lamellas move perpendicular to the
glass façade (Fig. 4).
The preliminary design-to-factory-file process of the innovative façade system was
simulated and a 3-D mock-up tested by the Fraunhofer Institute of Solar Energy (ISE) and the
special façade contractor Frener and Reifer from South Tyrol in Italy under the supervision of
General Planner ECE and the client ThyssenKrupp Real Estate. Then the different analog and
digital file formats for the Q1 value chain control batch of 22,000 lamellas were tested, cleaned,
calibrated, and digitally processed by the Simultaneous System Engineering group of
ThyssenKrupp Umformtechnik and ThyssenKrupp Nirosta, the group’s automotive
manufacturing units.
In general, the ThyssenKrupp System Engineering workflow contains an integrated
project process and system management for automotive systems and parts. This includes all
project phases beginning with the parametric concept study through to production design,
physical and digital mock-up (DMU) testing, assembly, and production planning—in this case—
to the façade system and component delivery for Q1. The integration of testing within the early
phase of the development is predominantly important Frener+Reifer, Thyssen Krupp
Umformtechnik, and Nirosta used for -controlled just-in-time (JIT) supply chain production such
as the integrated KANBAN supermarket system and optimization protocols with RFID radio
frequency identification (RFID).
The programming of the sensor-driven control system includes real-time daylight
provision in the interior, glare control, seasonal sun position, and the weather patterns assisted
through the weather station on the roof. The fully automated façade system is separated with a
parallel the logistics from fabrication, and transportation to assembly a barcode for the
correlation of multidimensional simulation models through finite element methods and
contributes to the verification of the computation results before any fabrication planning takes
place and bus system from the overall Siemens BAS and energy control system, coupled in a
central control gate to minimize the risk of collisions of the sun protection lamella components
and potential data package overloads that could occur with direct decentralize data package
integration into the overall BAS control system. Based on the previous algorithmically
© ASCE
developeed scenario simulations

s and
a detail reecommendattions for the sensor baseeline protocools by
the Fraunnhofer Instittute, all the slats of the façade systeem turn outwwards so thaat the sun shhades
remain open
o on clou udy days. Even
E when the
t slats aree closed direectly in fronnt of the faaçade,
employeees can open n the window ws and acceess for mainntenance is always posssible. The faaçade
contracto
or Frener an nd Reifer in nstalled the inner curtaain wall wiith German Schüco Syystem
elements. Both skins in Q1, curtaain wall, and d fins are appproximately 8,000 m² eaach.
TOTAL GREEN BUILDING

B AUTOMA
ATION SYS
STEM WIT
TH HUMAN
N-COMPUT
TER-
INTERF
FACE (HCII)
The
T German n Thyssen-KKrupp headq quarter’s reaal-time SIEM MENS totaal green buiilding
automation system (BAS) performs with intelligent control feeedback looops and leaarning
algorithm
ms for constaantly optimized building g performannce, security,, and user coomfort operaation.
This systtem also inclludes a wireeless environ
nmental mannagement sysstem to ensuure trend anaalysis
and optiimizations toward yeaarly mandattory net-zerro-energy ccertificationss based onn the
mandatorry yearly European Bu uilding Enerrgy Performmance Certifi fication systeem. The BA AS is
centralizeed, interlinkked, and a sensor driiven humann-computer-iinterface (H HCI) networrk of
hardwaree and softwaare. The leaarning autom mation systeem ensures the operatioonal perform mance
(transporrtation, light, water, HVAAC, energy , storage annd distributioon, etc.) of thhe Q1-8 faciilities
as well as the comforrt and safety
y of building occupants. TThe BAS opperates with intelligent aagents
(IAs) annd machine learning algorithms
a by
b identifyinng patterns for real-tiime optimizzation
potential including time
t schedulling and trend logging and verificaation of buillding autom mation
process.
Figure 4:
4 Room au or solar shading, dayligght control, artificial ligghting, heatting,
utomation fo
and cooling,
c dew
w point control, benchm
marked with h Energy Effficiency Claassification A
accoording to th
he BACS Eu uropean Norrm (EN) 152232. Diagraam adapted by Thomass
Spiegellhalter. (Sou
urce: Huberrt Dierkes, Siemens
S Inffrastructuree & Cities S Sector Build ding
Technoloogies Division, Essen, G
Germany, 2013.)
© ASCE
Siemens uses wireleess, automatted, self-learrning two-pposition algoorithm sensoor infrastrucctures

that consstantly contrrol and fine--tune buildinng spaces annd zoning cconditioning demand. Tooday,
fully inteegrated multidimensionaal trend datta processingg allows eff ffortless evennt-driven poolling
and analy ysis of real--time (onlinee) data and (offline)
( histtorical data in compliannce with muultiple
standardss. Any energy/water /reesource use and cost repports includding CO2 orr net-zero-ennergy
building (Net-ZEB) benchmarkiing values are a assemblled and pollled in real-ttime at any time
during th
he operation of buildingss.
Figure 5: Overa
all systems topology of the
t buildingg services w
with calculatting and selff-
learnin
ng algorithm
ms. Diagramm adapted an nd translateed by Thom
mas Spiegelhhalter. (Sourrce:
Hubert Dieerkes, Siemens Infrastrructure & C
Cities, Essen
n, Germanyy, 2013.)
BENCH
HMARKINGG RESULTTS THROU
UGH THE
E Q1 TOT
TAL GREE
EN BUILD
DING
AUTOM
MATION (D
DESIGO)
Siemens speccifically deveeloped a BA AS called Grreen Building Monitor™ ™ to supervisse the

building consumption n data of Q1
1 continuoussly for experrt analysis. T
The BAS inccludes a speccially
developeed energy-effficient operaating controol concept foor 1,133 roooms. Sensorss record dayylight
and deteect the preseence of peo ople, which allows the system to provide opptimal worksspace
conditionns based on those
t readin
ngs. The systtem displayss accurately to the hour hhow much ppower
and wateer is consumeed in the buiilding, the teemperature, and greenhoouse gas emiissions. The BAS
monitors user behav vior and com mpares, meaasures, and visualizes rreal-time prrogress data with
various performance
p benchmarkss for carbon neutrality oor the nearly--zero fossil eenergy standdards.
Other feaatures includde integrated control fo or electrical and securitty technologgy tailored tto the
specific needs
n of the ThyssenKru upp campus buildings (F Figure. 5).
LUSION
CONCL
The
T author iss still in neg
gotiation witth SIEMENS
S and Thysssenkrupp to receive the three
years (2
2012-2015) real-time operating
o Q1
Q Total G Green Buildiing Automaation (DESIGO)
© ASCE
benchmarking data. Unfortunately, the benchmarked real-time data was not yet permitted for
release. It would have helped to analyze the efficiency of how much carbon has been reduced
and energy costs for revenue growth with return on capital employed (ROCE). These pending
results will definitely feed into a future research case study with new post-occupancy interviews
and benchmarking.
In summary, one can argue that already today and in the future, automated green building
design-to-factory-file manufacturing and building performance automation will go naturally
together with faster and more flexible customization and corporate sustain ability strategies to
reduce carbon footprints and energy costs for revenue growth with return on capital employed
(ROCE). Over the next couple of decades, we will see major enhancements in automated
scenario network planning and in high-speed cloud computing that will further improve resource
innovations and flexibility. On the side of architects and engineers the next generation will
design and operate within a type of Green Automation, where designers and engineers deal with
graphical descriptions of system and complex cloud software with machine learning algorithms
that automatically translate repeatedly new models into optimized executable software. This new
era of green automated design to-virtual-to-real manufacturing, building operation and
automated benchmarking and scenario reporting will reorder the global AEC business for
decades.
REFERENCES
Frener&Reifer (2015). http://www.frener-reifer.com/projekte/thyssenkruppquartier-hauptver-

waltung/ accessed on March. 15, 2015.
SIEMENS DESIGO Building Automation, (2015). www.buildingtechnologies.siemens.com/
bt/global/en/buildingautomation-hvac/building-automation/building-automation-and-
control-system-europedesigo/Pages/desigo.aspx. Accessed on March. 15, 2013.
Institute for Human Factors and Technology Management (2015) of the University of Stuttgart,
Germany. http://www.iao.fraunhofer.de/lang-en/, accessed on March 23, 2015.
Interview with Hubert Dierkes (2013), Siemens AG, Infrastructure &Cities Sector Building
Technologies Division, Germany, on June 15, 2013.
Interview with Juergen Steffens (2013), JSWD Architekten, Cologne, Germany, on March 13,
2013.
Interview with Prof. Dr. Guenther Wenzel (2013), FRAUNHOFER IAO (Institut für
Arbeitswissenschaft und Technologiemanagement IAT der Universität Stuttgart) on March
18, 2013.
Thomas Spiegelhalter (2013). Mitigate, Adapt, Sustain: Emerging Workflows and Design
Protocols for Subtropical Carbon-Neutral H2 Cities, 2013, ACSA FALL CONFERENCE
DESIGN INTERVENTIONS FOR CHANGING CLIMATES
ThyssenKrupp (2011). “ThyssenKrupp Magazine—Architecture. www.thyssenkrupp.com.
accessed on March 23, 2015.
ThyssenKrupp (2011). “ThyssenKrupp Quarter: JSWD Architekten Chaix & Morel et Associés.”
Jovis; Bilingual edition (April 30, 2011).
© ASCE
Green Building Design as If People Mattered

Maryam H. Kashani1; Lyn Bartram1; and Robert Woodbury1
1
Human Centered Systems for Sustainable Living, School of Interactive
Arts+Technology, Simon Fraser University, P.O. Box V3T 0A3, Surrey, BC, Canada.
E-mail: mhaghigh@sfu.ca; lyn@sfu.ca; robw@sfu.ca
Abstract
Sustainability and energy efficiency are crucial to many sectors and all
stakeholders at local, municipal, national, and global scales. According to the city of
Vancouver, buildings contribute more than 50 percent of the city’s total GHG
emissions. With the introduction of standards such as LEED the hope has been
significant reductions in the footprint; however, change has been much lower than
predicted. Much of this can be attributed to how people actually use the buildings
they inhabit. Preliminary results of semi-structured interviews with residents of a
Vancouver multi-residential LEED Gold building show lack of feedback the most
common barrier among inhabitants who intend to save energy. Lack of appropriate
affordances and constraints follow respectively.
INTRODUCTION
Conservation and efficient use of energy are now major items for discussion
and action by governments, public companies, private groups, and individual people.
Various technical and non-technical sustainable solutions have been proposed at both
macro scales (urban fabric), and micro scales (individual buildings). However, it is
mostly only in the design of the urban fabric that we see evidence of people involved
in the design process (Haas, 2012). In the case of individual buildings, strategies
ranging from passive, active, renewable energies, renewable and non-renewable
materials, efficient building management system, and water and waste management
have been researched and discussed in detail (Fuad-Luke, 2002; Kibert, 2012;
Lechner, 2009). What is lacking in all these approaches is the effects buildings have
on the people using them, and, conversely, the effects of inhabitants on the built
environment. Furthermore, there is an extremely limited amount of literature that
looks at sustainable, green building design with the lens of inhabitants as the main
focus (Nielsen & E-Types (Firm), 2010; Strengers, 2013).
It is useful to think of buildings in terms of how people can affect them and
visa versa. Though this notion might seem complex and abstract, it is evident in
architectural history. Taking a careful look at the development of architecture and the
built environment throughout the years, we will see that as the interpretation and
perception of life changes, architecture has been transformed. The notion of
‘Functionalism’, which first appeared in 1932 in a letter addressed to Alberto Sartori,
was meant as a substitute to ‘rational’. Later, the meaning of functionalism morphed
© ASCE
to suggest that user behaviour can be calculated through the equation of

behaviour=function+form. This implies that the proportion of a space, arrangement
of rooms, and the way a house is located are all “expressions of the architect’s
thoughts on how life should be lived, and therefore ultimately his conception of what
the users should or should not be thinking and feeling” (Nielsen & E-Types (Firm),
2010, p. 57). One can extend this notion further into engineering, technologies,
controls and systems in the built environment where the location, shape, and
arrangement are all reflections on how they were intended to be used. However, the
major problem with intentions is precisely that they are intentions, which may have
little correspondence to what actually happens when a technology is used. In contrast,
this research takes a contextual, system-oriented approach. Every intention and
behaviour should be studied in the context of the system in which it occurs. Each of
the designed elements of a system may be able to be modified for efficient function
within these contextual circumstances. This approach is essentially that of human
computer interaction (HCI) (figure 1). The field of HCI is multi-disciplinary by
nature and influenced by social sciences (lifestyle and behaviour which are
contextual), engineering, and design (which take a system oriented approach).
Research Goal & Research Questions. Our goal is to view current frameworks of
green architecture and engineering through the lens of human-computer interaction
(HCI) in order to define new approaches to sustainable design thinking for architects.
HCI focuses on how people use technological systems and information and the
implications of designing and deploying those systems; buildings themselves are
complex combinations of systems. We re-frame the concepts of usability and design
from the HCI field to building design and propose a new framework for architectural
design interventions where our re-framing provides the main lens for both analyzing
sustainable architecture and providing guidelines and methods for architects to better
design for effective inhabitant use.
Our general research questions are: How can we devise and apply a new
interaction design lens and methods to the current green building frameworks to
augment their performance? What can designers and architects of the built
environment do differently in their process and approach to design?
In this short paper we start by describing the key principles and frameworks
for the study, followed by details of the semi-structured open-ended interviews, a
discussion on the results of the study, and finally a brief on future work.
BACKGROUND & DEFINITIONS
This paper aims at bringing a human centered lens to the field of green
architecture and engineering by reframing the way in which these two fields are
connected. Hence, we propose figure 2 by adapting figure 1 in the context of green
building design. Here, we identify specific concepts from HCI that are particularly
applicable. Similarly, from the social sciences, we identify key concepts for
understanding behavior in the context of green buildings for enhanced energy
efficiency. The largest change combines the nodes of engineering and design into a
single node, structured by the building artifact itself, not by the decision-making
© ASCE
structure by which it comes to be. The following section elaborates concepts from the
now three areas: HCI, social sciences and buildings.
Social Sciences
(2) Lifestyle Factors
Barriers
Social HCI
(1) Sciences Design Principles →

Design Opportunities
HCI Engineering Affordance

Engineering & Design →
Feedback Civil, Electrical, Mechanical
& Architectural Design
Design Constrains
v Site
Structure
Skin
Service
Space
Stuff
Figures 1 & 2 from left to right: (1) Human Computer Interaction draws from
social sciences as well as engineering and design disciplines (Shneiderman, 1992).
(2) Reframing figure 1 to account for architecture and engineering specifically in
the realm of energy saving.
HCI Common Engineering & Design Principles. In the widely cited popular book
of The Design of Everyday Things, Norman identifies seven design principles that
should guide every object we interact with in our lives (Norman, 2002). The
principles defined relevant to the field of HCI are:
1. Affordances: This concept provides a different perspective to the design of
environments and objects. It highlights the relationship between design intentions
and the user. For example, the form of an object or building and the behaviour of
the user/occupant as these link in practice (Maier, Fadel, & Battisto, 2009). For
example, a glass can be for seeing through and breaking. Adding the notion of
breaking to the actual function of glass is part of its affordance. This principle can
be essentially described as the functions or possibilities that are offered to the end
user through design.
2. Feedback: Systems should be able to give instant feedback that is “immediate and
synchronized with the user action” (Kai, 2010, p. 1). For example, playing a piano
provides immediate sound and key “feel”. There is a huge body of research on the
cognitive basis of feedback in interactive systems (Cialdini, 1993; Fogg, 2002).
3. Constrain: Lacing restrictions to some interactions can reduce the chance of mis-
use or errors. If we combine constraints with persuasion and feedback and make
sure each approach is selected based on the user’s behaviour at that point of time;
we can achieve more than either alone (Norman, 2002).
Barriers to Energy Saving. Lifestyle and living conditions affect how each person
faces barriers and impediments to saving energy. There are impediments that can be
modulated by small interventions, while other barriers need change at a larger level,
© ASCE
for example th hrough policiies. Althoug gh the latter has demonsstrated influuence in the
red
duction of en nergy consu umption, thee former hoolds significaant value, ssince it can
happpen in a shoorter time fraame and may y be easier tto achieve. BBased on an online semi
stru
uctured survvey conducted among 274 2 people residing in British Collumbia, we
disccovered the most domin nant barrier to
t energy saaving is lackk of informattion or mis-
info
ormation, foollowed by cost,
c attitudee and preferrence, technnical and spaatial design
bassed barriers, living condiitions and lasstly perceiveed lack of aggency.
Arcchitectural Design Fra amework. A very comm mon framewoork for desiggn practices
fromm 1994 is to t look at buildings
b wiith the lens of what theey are madee and what
happpens to them m after they are construccted (Brand,, 1994). Builldings compprise several
layers, with wh hich the inh
habitants con nstantly interract and whhich change at different
ratees. In How Buildings
B Leearn, Brand discusses hoow buildinggs adapt to change. One
of the
t key argu uments he makes
m is in terms
t of thee different shhearing layeers (“shear”
refeers to the rellative longev
vity of buildiing componeents) in eachh building. H He proposes
thatt buildings can
c be seen as comprisiing six layerrs (figure 3).. He definess Site as the
settting and surrrounding off a structure,, Structure aas the founddation and looad bearing
elemments of a building, Skkin as the laayer that coovers the strructure, Servvices as the
commmunication n hub of the building, HVAC
H systemms, electricaal wiring, eleevators, etc,
Spaace plan ass the interio or layout off a buildingg, and Stuff ff as the furrniture and
belongings of occupants.
o
Figure
F 3: Bra
and's six sh
hearing layers of changee, and the n
number of yyears they
can exisst (Brand, 1994)
INH
HABITANT
T STUDY
Tallking to Reesidents. Baased on figuure 2 we carrried out a sstudy that eengaged the
inh
habitants of “green” buiildings, baseed on semi--structured, open-endedd interviews
witth 12 residen
nts of a LEE
ED Gold mullti-residentiaal apartment unit in Vancouver-BC.
Alll residents were
w owners.. As such, th
he lifestyle ffactors that iinfluenced ttheir energy
con
nsumption were
w similarr. We used the conceppts and structures of thhe previous
section in designing the interview questions.
q Inn particularr, we took a “human-
com nding these ppatterns: in essence, we considered
mputer interaaction” lens to understan
the inhabitant as a user and the buiilding as a complex syystem with which the
© ASCE
© ASCE
Elements of Opportunities- Opportunities- Opportunities- Barriers
Architecture Affordances Feedback Constraints
Site Location of building.1 No yard.11
bottom.
Structure
Skin Floor to ceiling windows.2 Immediate areas around the

Operable windows.3 large windows.21
Windows in the bathroom.22
Sustainable Human–Building Ecosystems
Brand’s six layers in table 1.
Service No garbage shoot.12 Lack of separate zones for

Bedroom power fuse heating and cooling in the
popping if too much home.23
electricity used at one point Extremely complex
of time.13 thermostat and settings. 24
You can’t get the
immediate area heating
unless you over heat the
unit.
Pre programmed bathroom
fan.25
Bad odour from bathroom
drainage.26
Space Size and layout of the Location of pot lights, and

space.4 light junction boxes.27
Table 1. Summary of study data
High ceiling.5 Placement of light switches

spatial design, alerts, and site, and opinions on sustainability.
in the space.28
Odd placement of air vent.29
Stuff Larger sink in kitchen.6 Beeping sound from dryer, Predetermined wash No feedback from the stove.
Dual Flush Toilet. dishwasher and cycles .14 Marble floors in the
Task lighting and light fridge/freezer.8 Smart Dryers.15 bathroom.30
controls.7 Controlling blinds.9 Smaller showerheads .16 Nice rain shower.31
Bedroom power fuse Smaller load washer Lack of dimmers on light
popping if too much /dryer.17 switches.32
electricity used at one point Smaller fridge.18 Lack of lights in the den
points are elaborated in the order of left to right, and within each column from top to
of time.10 Faucets with less water and bedrooms.33
Table 1 briefly summarizes the main points that came out of the study. Below the key
Results. We categorized data from discussions with residents using the specific
concepts from figure 2, and mapping the HCI and social science concepts against
inhabitant interacts. We discussed living experience, thermal comfort, lighting,
pressure.19
Door handles.20
LED Lights only
180
Opportunities - Affordances:
1
Building is located in an area with close proximity to transit and general services.
2
Floor to ceiling windows afford the transmission of light, and hence illumination of
the interior environment as well as a view of the exterior environment. “Even on
rainy days apartment feels cozy”.
3
Operable windows may afford the exchange of air.
4
Apartment size is much smaller than their large house.
5
High ceiling make the space look bigger, without having to actually add to the size
of the apartment.
6
Larger sink in kitchen allows for more water to fill up, hence more flexibility of
choice.
7
Toilet affords dual flush for less water use.
Opportunities - Feedback:
8
Beeping sound from dryer, dishwasher and fridge/freezer.
9
Controlling the blinds gives you direct experience of something that can indirectly
control energy use. But it affords privacy.
10
Bedroom power fuse popping every half an hour if all plug outlets used at the same
time (ie, working from home and plugging in computers and printers, or electric
heaters).
Opportunities - Constraints:
11
No yard means less maintenance and water use.
12
Lack of garbage shoot causing more recycling habits, because neighbours might see
what each household is throwing out in their trash bags.
13
Bedroom power fuse popping every half an hour.
14
Predetermined wash cycles afford use of less water and energy.
15
Smart dryers stop when detecting the clothes are dry, affording less energy use.
16
Smaller showerheads allow less water out in each unit of time.
17
Smaller load capacity in washer/ drier allows frequent washing with less waste.
18
Smaller fridge means less groceries, more frequent shopping, and less rotting
vegetables and waste.
19
Faucets are water savers cause they have less pressure.
20
Door handles that do not afford the understanding of whether they should be pushed
or pulled, hence leaving them open at all times. This way there is visibility of whether
any lights have been left on when leaving the apartment.
Barriers:
21
Immediate areas around the large windows feel cold in winter and warm in summer.
22
Windows in the bathroom, make bathrooms very cold, hence residents turn heat up.
23
Lack of separate zones for heating and cooling in the home cause residents to
heat/cool up all the apartment even if they only intend to heat/cool up one room/space.
24
Extremely complex thermostat and settings cause mis-use of the thermostat.
25
Pre programmed bathroom fan, which is not visible but automatically turns on and
off creates confusion and extra waste of electricity since inhabitants also turn the fan
on and off when required.
26
Bathroom drainage has a bad odor coming out of it causing residents to constantly
pour boiling water in it, hence wasting hot water.
© ASCE
27
Pot lights are on the perimeter of the living room, and light junction boxes in some
places but never in the right location. So, all the lights have to be on, or furniture has
to be moved around to fit the junction box.
28
Placement of light switches in the hallway leading to bedrooms is way too far from
the bedroom itself.
29
Odd placement of air vent, giving hot air in summer and cold air in winter
constrains the possibility of sitting beside the window.
30
Marble floors in bathrooms feel cold to walk on causing residents to increase heat.
31
In some bathrooms with nice rain showers inhabitants tend to take longer showers.
32
Lack of dimmers on light switches causes one to waste more electricity.
33
Lack of light in the bedroom and den causes residents to try to brighten up these
rooms through turning on the light in other areas of the home excessively.
DISCUSSION
Results from talking to residents indicate that inhabitants perceive that the
structure layer has little influence. Though the LEED point sheet in this building
clearly indicated the skin and the site to have obtained the highest points, inhabitants
saw constraints in stuff to be the most dominant opportunity while saving energy.
This once again refers back to our initial goal to focus on the way inhabitants interact
with their surroundings, because despite the number of points a green building might
score, it ultimately comes down to the actions inhabitants take. In addition to the
categorical insights from residents we have included a set of direct quotes below:
Inhabitant#2: “LEED means luxury. It is a buzzword that adds to the resale value”.
Inhabitant#3: “I never got to see where the LEED standard is happening. How can I
know it is LEED?”. Inhabitant#6: “If LEED is what you say, then the building is
taking care or energy saving itself. Energy saving is automatically in place”.
Inhabitant#8: “In general I don’t think about energy saving. If I saw the LEED thing
it would be a great learning tool for both kids, and adults. I absolutely would love to
have a list of LEED points”.
These comments all point to the fact that lack of information is the most
dominant barrier inhabitants face to saving energy. Though this wasn’t initially part
of our reframed notion of architectural design, but it provided additional valuable
insight that can be useful for the multiple players involved in the design and
execution of green buildings.
CONCLUSION & FUTURE WORK
Considering the interdisciplinary nature of architecture and the goal of

applying the HCI design principles to this field we summarize insights of our study
based on a specific context, while accounting for behavioural insights from residents.
Our hypothesis is that the HCI lens can give new views on more than just inhabitant
experience. It may give us new tools for design.
A format widely used in human–computer interaction (HCI), primarily in
interface and web design, is that of the design pattern, which describes a form of
presenting a situation, and/or possible solutions, in a structured way. “Patterns are
© ASCE
essentially recurring problem-solution instances, described in a referenceable way

which enables practitioners to recognize the situation” (Lockton, Harrison, & Stanton,
2008, p. 434). Two widely used refinements of the design pattern in HCI design
include personas and scenarios, providing a means to explore various design
approaches in different contexts (Bartram & Woodbury, 2011). The form, via
adoption in software engineering in the late 1980s, stems from architecture:
(Alexander, Ishikawa, & Silverstein, 1977). ‘Pattern Languages’ summarized
concisely prior professional discourse, which covers the design and layout of
buildings, towns and communities. The pattern form can help a designer recognize
that a ‘new’ problem situation is similar or analogous to one encountered (and solved)
previously elsewhere, even in a different context. Sheets such as the LEED rating
system, and sustainable design principles, etc. are all forms of design patterns. We
hope that our HCI lens becomes an additional element to the current design patterns
suggested in the sustainable design industry, one that focuses on inhabitants, their
lifestyles, and behaviour.
We aim that, by combining concepts and tools from HCI and
design/engineering we can develop new ways of effectively using results such as
those reported here in professional practice. To this end, we are currently designing a
series of workshops with designers and engineers with the specific aim of discovering
how to better combine inhabitants’ views, behaviour and experienced barriers with
the affordances, feedback and constraints of design and engineering.
REFERENCES
Alexander, C., Ishikawa, S., & Silverstein, M. (1977). Pattern languages. Center for
Environmental Structure, 2.
Bartram, L., & Woodbury, R. (2011). Smart Homes or Smart Occupants? Reframing
Computational Design Models for the Green Home. Proceedings of the AAAI
Spring Symposium on Artifical Intelligence and Sustainable Design, 1–8.
Brand, S. (1994). How buildings learn: what happens after they’re built.
Cialdini, R. B. (1993). Influence: The psychology of persuasion. New York, Morrow.
Fogg, B. J. (2002). Persuasive technology: using computers to change what we think
and do. Ubiquity, 2002(December), 5.
Fuad-Luke, A. (2002). The Eco-design handbook. Thames & Hudson London.
Haas, T. (2012). Sustainable Urbanism and Beyond: Rethinking Cities for the Future.
Kai, C. (2010). Design Principle for Ubiquitous Computing.
Kibert, C. J. (2012). Sustainable construction: green building design and delivery.
Wiley. com.
Lechner, N. (2009). Heating, cooling, lighting: sustainable design methods for
architects. John wiley & sons.
Lockton, D., Harrison, D., & Stanton, N. (2008). Making the user more efficient:
Design for sustainable behaviour. International Journal of Sustainable
Engineering, 1(1), 3–8.
Maier, J. R., Fadel, G. M., & Battisto, D. G. (2009). An affordance-based approach to
architectural theory, design, and practice. Design Studies, 30(4), 393–414.
© ASCE
Nielsen, K. H., & E-Types (Firm), D. arkitekturcenter. (2010). Mind your behaviour:
how architecture shapes behaviour. [Copenhagen]: 3XN.
Norman, D. A. (2002). The design of everyday things. Basic books.
Shneiderman, B. (1992). Designing the user interface: strategies for effective human-
computer interaction (Vol. 3). Addison-Wesley Reading, MA.
Strengers, Y. (2013). Smart Energy Technologies in Everyday Life: Smart Utopia?.
Palgrave Macmillan.
© ASCE
Integration of QFD and Utility Theory to Improve End-User Satisfaction in

the Design of High-Performance Buildings
Ehsan Mostavi1; Somayeh Asadi1; Ebrahim Karan2; and Djamel Boussaa3

1
Department of Architectural Engineering, Pennsylvania State University, 104
Engineering, Unit A, University Park, PA 16802. E-mail: asadi@engr.psu.edu
2
Department of Applied Engineering, Safety & Technology, Millersville
University, P.O. Box 1002, Millersville, PA 17551-0302.
3
Department of Architecture and Urban Planning, Qatar University, P.O. Box
2713, Doha, Qatar.
Abstract
Ever since the first global oil crisis in the seventies, the shortage of fossil
fuels, which is considered as the number one resource for energy carrier, has
emphasized the need of modern society on cheap energy and resources. Over
short or long term, that circumstance is compelling people to consider global
energy strategies and accordingly take proper measures. Along with a limited
supply of carbon based fuels worldwide, the impact of the carbon footprint— the
effect of an increasing carbon dioxide (CO2) concentration in the earth’s
atmosphere and its consequence on the global climate — has become indubitably
clear. Recent studies show that building sector consumes significant amounts of
energy and is responsible for a considerable amount of greenhouse gas (GHG)
emissions. Studies show that amongst several factors contributing to the building
energy performance, occupant behavior is one of the driving factors. According to
these studies building characteristics and the quality of indoor environment
significantly influence the level of satisfaction and efficiency of occupants in
buildings. Therefore, a proper design is needed to increase building occupant’s
satisfaction and efficiency.
To determine the best design strategy, several building envelope materials
as well as occupant behaviors should be considered as design variables in the
design process. Therefore, the objective of this study is to develop a proper
weighting system for different strategies to secure occupants’ satisfaction as well
as constructors’ consent. To achieve this objective, a Quality Function
Deployment (QFD) method is utilized to identify, translate, and implement the
end-user requirements into the design process. In addition, the utility theory is
used to reflect the constructors’ perception on usefulness and practicality of end
user requirements. The insights from this study will aid in developing better
decision making system and more viable alternatives.
Keywords: Energy consumption; Occupants’ satisfaction; Quality function

deployment; Utility theory.
© ASCE
INTRODUCTION
Ever since the first global oil crisis in the seventies, the shortage of fossil
fuels has emphasized energy efficiency in industries such as construction and
transportation and the use of cheap energy and resources. Along with a limited
supply of carbon based fuels worldwide, the impact of the carbon footprint (i.e.
the effect of an increasing carbon dioxide (CO2) concentration in the earth’s
atmosphere and its consequence on the global climate) has become indubitably
clear. Recent studies show that building sector consumes significant amounts of
energy and is responsible for a considerable amount of greenhouse gas (GHG)
emissions. According to the United Nations (UN), the building sector consumes
40% of global energy consumption and has a crucial role in mitigating the impacts
and risks of global warming [1] .
Optimal energy-efficient building design should consider multiple
objectives (e.g. energy consumption, financial costs and environmental impacts)
and thus multi-objective optimization is a promising direction to follow. There are
numerous optimization methods to find the tradeoff between different design
variables and optimum strategy for construction of an energy efficient, low cost,
and environmentally sustainable building [2][3][4][5][6]. In a study performed by
Weimin [7], the environmental impacts of buildings were investigated. The
objective of this study was to minimize the life cycle cost (LCC) and life cycle
environmental impact (LCEI). To achieve the objective, the genetic algorithm was
used to develop the multi objective model. They demonstrated that the utility
structure has a significant impact on the building’s environmental performance. In
other study conducted by Hamdy [6], a three-phase genetic algorithm multi-
objective optimization model (PA-GA-RF) was developed and utilized to
minimize the CO2 emission and investment cost of a two story building and its
HVAC system.. It was found that the optimal combinations or Pareto front could
reduce the carbon dioxide equivalent (CO2-eq) emission by 32% and initial
investment by 26%. In another study, Fesanghary et al. [8] developed a multi
objective optimization based on harmony search algorithm (HSA) to minimize the
LCC and CO2-eq. They considered a number of different building envelops as
design variable parameters in optimization process. A comprehensive life cycle
assessment including pre-use (extraction, raw material processing, transportation,
and construction), use (all emissions during building life), and end-of-life
(demolishing, and landfilling) was considered in this study. According to defined
model, authors suggested series of solutions for economically and
environmentally efficient design.
While these optimization models are capable of determining the best
construction strategy, implementing the end-user satisfaction preferences into the
optimization process is still a missing puzzle piece. Satisfaction metrics — the
state of fulfilment—are generally qualitative and depend on many different
factors. They can be recorded by short surveys and questionnaires of participants’
attitudes toward a given design or issue. For example, Brokelman et al.[9], applied
visual analogue scale (VAS) to assess patients’ satisfaction in the health care
system. In this method, patients’ satisfaction rates are marked on vertical line with
length of 100 units. VAS is the horizontal line having two descriptors
demonstrating extreme satisfaction levels (lowest and highest satisfaction level) at
both ends. The validation performed by measuring Spearman correlation test
which showed that pain is the most important patient satisfaction factor. In
© ASCE
another study, Castillo et al. [10] identified and modeled users’ satisfaction of
public transportation system. Satisfaction level of different parameters of
transportation system comfort were determined through a survey asking
individuals to rate their satisfaction level. Three models including average based
model, multivariate discrete distribution and a generalized linear model were used
to model the overall satisfaction level of individuals as a function of specific
satisfactions.
There are numerous clear evidences denoting that building characteristics
and the quality of indoor environment significantly influence the level of
satisfaction and efficiency of occupants in buildings. Therefore, to fill this gap and
increase the practicality and satisfactory level of the proposed strategy the
objective of this study is to develop a proper weighting system for different
possible strategies to assure occupants’ satisfaction as well as constructors’
consent.
METHODOLOGY
In this study, possible strategies which could be proposed by occupants are

weighted and evaluated based on their occupants’ satisfactory and practicality. To
achieve the objective of creating the weighting system, customers’ preferences are
determined and evaluated using Quality Function Deployment (QFD) method. In
addition, the utility theory is used to reflect the constructors’ perception on
usefulness and practicality of end user requirements. The first part of the study
aims at investigating the factors influencing occupant satisfaction and behavior
through literature review and semi-structured interviews of individuals and
building occupancies. Next, the constructors’ perception on usefulness and
practicality of these factors will be measured by using utility theory.
House of Quality. House of quality (HOQ) is a diagram used to illustrate the

relationship between customers’ requirements and producer’s capabilities. HOQ is
a part of Quality Function Deployment (QFD) and consists of 6 major parts: 1.
Customer requirements which are usually established through interviews with end
users, 2. Planning matrix (on the right side of the HOQ) which quantifies the
customers’ priorities, 3. Technical requirements box which are determined
through interviews by constructors (producers), 4. Interrelations box that shows
the level of relation between customers and technical requirements, 5. Roof used
to identify the effect of different technical requirement on each other, and 6.
Target box that summarizes the conclusions drawn from the entire matrix. In this
study HOQ is built by a comprehensive interview with graduate students working
in offices as well as office staffs. A separate set of interviews was also conducted
to determine the construction experts’ capabilities and concerns. The involvement
of 24 college students and staffs consisted of the participation in interviews aimed
at identifying 25 factors influencing the occupants’ attitudes and satisfaction level.
Utility Theory. As buildings occupants’ satisfaction becomes more crucial in

construction industry, many studies have tried to regulate construction industries’
performances which can affect the occupants’ satisfaction. Under the recent
special attention to increase the overall occupants’ satisfaction, a parallel trend of
studies have suggested to evaluate the frameworks with various occupants’
satisfaction parameters enabling companies to quantitatively determine the level
© ASCE
of satisfaction of their performance. Despite the paybacks of using these

frameworks, it is difficult to select unit framework or set of parameters in terms of
usefulness and practicality. Generally the frameworks are created and suggested
by the optimization algorithms are not capable of considering the feasibility and
practicality of different strategies and eventually expert people in construction
industry are rarely utilizes the benefits of these studies and optimizations
suggestions. This section aims to evaluate the total utility of various occupants’
satisfaction parameters, which have been determined through the interviews
during HOQ creation focusing on usefulness and practicality.
As a company intends to apply satisfaction parameters in their project,
they have to consider how much practical and effective the parameter is. Hence,
identifying parameters with higher utility is necessary to propose the best
framework which is more practical and effective. In this section, the utility theory
approach was utilized to calculate the overall utility of each parameter. These
overall utilities provide useful information of the most preferred parameters to
increase the occupant’s satisfaction effectively.
Generally, utility theory is a tool to determine a decision maker’s preferences and
weightings of different alternatives by reflecting risk preferences. In other words,
utility theory will assist to determine the construction companies’ attitude towards
risk under uncertainty.
Basic concepts of utility and the value of an uncertain gamble was initially
proposed by Bernoulli[11]. He described utility as an individual’s perception or
satisfaction which is a unique function for each individual. Utility can determine
the best preferred strategy according to decision maker preferences: Value
Preferences and Utility Preferences. Value preference evaluation is used to select
the best alternative under deterministic decision environment while utility
preference is utilized to select between alternative with some levels of
uncertainty[12].
The following formula is used to derive the single attribute utility
function:
( )= − ×
Where A and B are scaling factors for the utility function and RT is risk tolerance.
A and B are functions of RT and the highest and lowest utility:
= ÷ −
=1÷ −
RT can be estimated by the following equation:
( )= 0.5 × + 0.5 × × (− )
The single utility function can be easily derived if the RT value is known.
The RT value can be computed from the certainty equivalent (CE) by inversely
calculating the RT value from the CE. Thus, the first step to obtain the single
attribute function can be the determination of the CE (which is equal to 3) for
each attribute. The CE value of each attribute is decided by the decision problem
such that the decision maker is indifferent between the CE and a lottery between
the best and worst in the corresponding attribute. The detailed procedure and
calculation is shown in the result section.
In the effort to create a useful and practical strategy to increase occupants’
satisfaction, 25 different parameters having the most influence on the occupants’
© ASCE
satisfaction level were determined. To determine the construction industry attitude

toward usefulness and practicality of these parameters, construction experts were
asked to fill two different surveys asking their evaluation of usefulness and
practicality of determined parameters. Experts were asked to grade the utility of
different parameters from 1 to 5 (5 the most useful/practical). Total number of 32
surveys were collected. The total utility of each parameter is calculated by
averaging the total 32 scores obtained from all the experts.
RESULTS
HOQ. As it can be seen in Table 1, 25 parameters which were selected through a

comprehensive interview with 24 grad students and office staffs were categorized
into 6 groups – Indoor Air Quality, Available Space, Accessibility, Available
Equipment, Furniture, and Textures. These parameters are considered to be the
end user satisfying requirements.
Table 1. 25 most influencing parameters on occupants’ satisfaction

Performance Technical Details Material and Range
No. of Floor Texture materials
No. of Wall texture material

Temperature fluctuation
Building Area Needed

○ - High (9)
Quality of furniture
□ - Medium (3) Amount of wiring
No. of Equipment
Temp Insulation
∆ - Low (1)
Meet standards
AC insulation
Having windows □ ∆ ○ ○
Enough lighting □ ○ □
Individual desk lamps □ ○ □
Proper Temperature
IAQ
(Summer) □ ∆ □ ∆
Proper Temperature (Winter) □ ∆ □ ∆
Noise Level (Instruments) ∆ ○ □
Noise Level (Coworkers) ∆ ○
Noise Level (outdoor source) □ ○
Having Private Space ○ ∆ ∆ ∆ ∆
Spaces
Enough Space for storing ○

Proper Lounge □ ○ □ □ □
Smoking Room ○ ∆
Plugin ○
Access
Switches ○
Temperature Adjusting ○ ○
Wireless Internet □ □
Equipment
Printer ∆ □
Copy machine ∆ □
Scanner ∆ □
Desks ○
Furniture
Chairs ○
Bookshelves ○
Indoor plants □
Texture
Flooring texture ○
wall texture ○
© ASCE
Utility Function. Utilities regarding to usefulness and practicality of parameters

which were determined in the previous section were calculated using mentioned
equation. Based on the average of usefulness and practicality level for each of the
parameters, following utility functions are calculated:
( ) = 4.325 − 7.302 × .
( ) = −0.758 + 0.193 × .
Table 2 shows the effectiveness and practicality utility of each parameter.
As it can be seen in this table, proper temperature in summers and winters are
having the highest effectiveness utility and providing offices with better chairs
and desks are having the highest practicality utility. On the other hand providing
enough storing space and having access to copy machine are having the least
effectiveness utility while having windows in offices and proper noise level due to
coworkers are having the least practicality utility.
Table 2. Usefulness and practicality of occupants satisfying parameters

Effectiveness Practicality
Parameter X U(x) X U(x)
1 Having windows in office 3.23 0.685 2.41 0.000
2 Enough lighting in office 3.05 0.541 3.75 0.866
3 Having individual desk lamps 3.02 0.516 3.21 0.436
4 Proper Temperature (Summer) 3.65 1.000 3.15 0.396
5 Proper Temperature (Winter) 3.48 0.876 3.24 0.457
6 Proper noise Level (Instruments) 3.02 0.516 2.87 0.227
7 Proper noise Level (Coworkers) 3.34 0.770 2.43 0.009
8 Proper noise Level (outdoor source) 3.25 0.701 2.89 0.238
9 Having private working spaces 2.77 0.306 2.65 0.111
10 Enough Space for storing 2.43 0.000 2.56 0.067
11 Proper lounge 2.89 0.408 2.89 0.238
12 Smoking room 2.90 0.417 2.72 0.146
13 Proper access to plugins 3.50 0.891 3.41 0.580
14 Proper access to lamp switches 2.45 0.019 3.55 0.691
15 Having access to temperature adjusting system 2.89 0.408 3.22 0.443
16 Proper wireless Internet access 3.35 0.778 3.61 0.741
17 Having access to printer 2.89 0.408 3.73 0.847
18 Having access to copy machine 2.47 0.037 3.20 0.430
19 Having access to scanner 2.66 0.209 3.65 0.776
20 Comfortable desk 2.96 0.467 3.85 0.960
21 Comfortable chairs 3.08 0.565 3.89 1.000
22 Having bookshelves in office 2.43 0.000 3.63 0.759
23 Having indoor plants 3.48 0.876 3.32 0.514
24 Type of flooring texture 3.33 0.763 3.50 0.650
25 Type of wall texture 3.12 0.598 3.59 0.724
Experts’ Attitude. Based on the given utility functions for usefulness and
practicality of occupants’ satisfaction parameters, the following utility graphs can
be plotted. As Figure 1 shows, constructors are having negative attitude
(downward curvature, risk averse attitude) toward usefulness of the proposed
© ASCE
parameters, however they have positive attitude (upward curvature, risk seeker
attitude) toward practicality of this parameters. Therefore utilizing the strategies
with higher utility of usefulness can be an effective motivation for selecting the
optimum strategy.
1
Utility 0.8
0.6
0.4
0.2
0
2 2.5 3 3.5 4
Usefulness
1
0.8
Utility
0.6
0.4
0.2
0
2 2.5 3 3.5 4
Practicality
Figure 1. Utility vs. Usefulness/Practicality
The list of top 15 parameters and associated rankings based on the constructors’
usefulness utility and associated assigned weighting are given in Table 3. These
weightings are calculated using the following equation:
( )
( ) = ∑
( )
Where W(X) is the weight of each strategy, U(X) is the utility associated to each
of the strategies.
Table 3. Usefulness of occupants satisfying parameters and relative

weightings
Parameter X U(X) W (X)
Proper Temperature (Summer) 3.65 1.000 0.078
Proper access to plugins 3.5 0.891 0.070
Proper Temperature (Winter) 3.48 0.876 0.069
Having indoor plants 3.48 0.876 0.069
Proper wireless Internet access 3.35 0.778 0.061
Proper noise Level (Coworkers) 3.34 0.770 0.060
Type of flooring texture 3.33 0.763 0.060
Proper noise Level (outdoor source) 3.25 0.701 0.055
Having windows in office 3.23 0.685 0.054
Type of wall texture 3.12 0.598 0.047
Comfortable chairs 3.08 0.565 0.044
Enough lighting in office 3.05 0.541 0.042
Having individual desk lamps 3.02 0.516 0.040
Proper noise Level (Instruments) 3.02 0.516 0.040
Comfortable desk 2.96 0.467 0.037
© ASCE
CONCLUSION
For the improvement of energy efficiency of the buildings in addition to

the satisfying the occupants’ requirements, decision makers are facing a multi
objective optimization. The optimization process should not only propose the
energy efficient combination with lower environmental footprint, but also the
proposed strategy should satisfy the occupants and the constructor has a positive
attitude toward it.
The aim of this paper was to develop of a weighting system involving end
users (occupants) and constructors into the decision making process. Weighting
system process consists of determining the parameters required by end user and
evaluating construction experts’ utility perception toward the required parameters.
End user satisfaction parameters were determined using QFD method and
constructors’ perception were evaluated via utility evaluation. It was seen that
constructors are having risk seeker attitude toward practicality of the strategies,
however their attitude toward usefulness of strategies are negative. The current
stage of the study however focuses on the limited parameters of satisfaction. More
parameters can be optimized and evaluated in further studies.
REFERENCES
[1] UN, “Building and Climate Change,” 2009.

[2] Y. Shao, P. Geyer, and W. Lang, “Integrating requirement analysis and
multi-objective optimization for office building energy retrofit strategies,”
Energy Build., vol. 82, pp. 356–368, Oct. 2014.
[3] C. Diakaki, E. Grigoroudis, N. Kabelis, D. Kolokotsa, K. Kalaitzakis, and
G. Stavrakakis, “A multi-objective decision model for the improvement of
energy efficiency in buildings,” Energy, vol. 35, no. 12, pp. 5483–5496,
Dec. 2010.
[4] C. J. Hopfe, G. L. M. Augenbroe, and J. L. M. Hensen, “Multi-criteria
decision making under uncertainty in building performance assessment,”
Build. Environ., vol. 69, pp. 81–90, Nov. 2013.
[5] S.-H. Lin and D. J. Gerber, “Evolutionary energy performance feedback for
design: Multidisciplinary design optimization and performance boundaries
for design decision support,” Energy Build., vol. 84, pp. 426–441, Dec.
2014.
[6] M. Hamdy, A. Hasan, and K. Siren, “Applying a multi-objective
optimization approach for Design of low-emission cost-effective
dwellings,” Build. Environ., vol. 46, no. 1, pp. 109–123, Jan. 2011.
[7] W. Wang, R. Zmeureanu, and H. Rivard, “Applying multi-objective
genetic algorithms in green building design optimization,” Build. Environ.,
vol. 40, no. 11, pp. 1512–1525, Nov. 2005.
[8] M. Fesanghary, S. Asadi, and Z. W. Geem, “Design of low-emission and
energy-efficient residential buildings using a multi-objective optimization
algorithm,” Build. Environ., vol. 49, pp. 245–250, Mar. 2012.
[9] R. B. Brokelman, D. Haverkamp, C. van Loon, A. Hol, A. van Kampen,
and R. Veth, “The validation of the visual analogue scale for patient
satisfaction after total hip arthroplasty,” Eur. Orthop. Traumatol., 2012.
[10] J. M. Del Castillo and F. G. Benitez, “A Methodology for Modeling and
Identifying Users Satisfaction Issues in Public Transport Systems Based on
© ASCE
Users Surveys,” Procedia - Soc. Behav. Sci., vol. 54, pp. 1104–1114, Oct.
2012.
[11] D. Bernoulli, “Exposition of a new theory on the measurement of risk,” J.
Econom. Soc., pp. 23–26, 1738.
[12] D. L. Thurston and A. Locascio, “Decision theory for design economics,”
Eng. Econ., 1994.
© ASCE
The Power of Data Visualization: A Prototype Energy Performance Map for

a University Campus
Khaled A. Tarabieh1; Islam O. Elnabarawy2; Islam A. Mashaly3; and Yussra M.

Rashed3
1
Assistant Professor of Sustainable Design, Construction and Architectural
Engineering Department, School of Sciences and Engineering, American
University in Cairo, AUC Ave., P.O. Box 74, New Cairo 11835, Egypt. E-mail:
ktarabieh@aucegypt.edu
2
Ph.D. Student, Computer Science Department, Missouri University of Science
and Technology, 1870 Miner Circle, Rolla, MO 65409. E-mail: ie3md@mst.edu
3
Research Assistant, Construction and Architecture Engineering Department,
School of Sciences and Engineering, American University in Cairo, AUC Ave.,
P.O. Box 74, New Cairo 11835, Egypt. E-mail: islammashaly@aucegypt.edu;
yussra.mer@aucegypt.edu
Abstract
A key factor to promote energy efficiency across campus and buildings is
through education and clarity of information. Energy data is complex to
understand and difficult to comprehend for the majority of users. Higher
education institutions consists of building groups that usually operate according to
strategies pre-set by the operation teams that respond to a specific energy demand
condition. The purpose of this research is to design a data friendly interface to
simplify and assist in the implementation of energy literacy across campus. The
research method utilized available energy metered and simulated data at the
campus level provided by the university operations. A web based data
visualization interface is developed and an energy map is created to inform the
campus community on the status of energy consumption. In addition to
visualization, the proposed interface helps operators to benchmark campus
buildings, promote energy efficiency across campus and disseminate the resulting
information in a simple, clear and concise way to the campus community. The
interface is a great step towards change of behavior through education, a
questionnaire to measure user actions and understanding is distributed to measure
the effectiveness of this new approach and the results are compiled to advance the
interface design and user experience.
INTRODUCTION
Based on recent findings in selected literature, clear building energy data

proved to be the key to making better energy-based decisions. Whether this data is
simulated or acquired through various meters, the usefulness of tracking it goes
beyond comparative analyses and benchmarking. Before data metering and
visualization became prominent, only simulated energy data could be utilized to
make better energy-based decisions regarding the different building components.
© ASCE
Malkawi et. al. developed a web-based user interface that employed simulated
thermal data of buildings to provide an energy-based decision tool for windows by
predicting its performance (Malkawi, 2007). However, as energy performance
change over the lifetime of the building, energy-based decisions concerning
systems and components should be included as well. Using only simulated data
for buildings will hardly give a real picture of how a building is actually
performing, hence in day-to-day operations, constant data feed and continuous
monitoring and analysis of the actual energy data is crucial to improve energy
performance and retain savings.
Domínguez et al. proposed a tool to collect energy data from university
buildings through data mining, monitoring and statistical techniques that led to
economic savings of around 15%. The three-layer structure of the developed
system started with acquiring data from campus meters, this data was stored and
processed then the client was able to access it online as a monitoring interface and
conduct a number of relevant statistical analyses (Domínguez, 2013). The
measured energy data from campus buildings was used to calibrate simulated
energy models. The calibration of simulated energy models often result in better
prediction for the building’s performance and accordingly better decisions were
made regarding the different building systems, therefore allowing for streamlined
calculation of the impacts was done with higher precision. For example, it was
possible to use collected real-time campus data, the way Dong et al. did, to
develop a building energy model and calibrate it to get performance predictions
within ±15% error band. Furthermore, the model was used to adjust energy
consuming such as the HVAC system and achieved a high savings rate of 39%
(Dong, 2014).
Recently, Yarbrough et al. utilized a university campus energy data
visualization and comparisons for a more specific analysis. The focus was
directed to the peak energy, tracking the buildings contribution and assessing the
type of activities and uses that add-up to that peak value (Yarbrough, 2015). In
addition, the monitored energy data was visualized and made into a tool for
comparative analyses ultimately to assist occupants’ with the required knowledge
that can lead to change in behavior and accordingly in saving even more energy.
A six week empirical study in a residential building on the campus of
Columbia University by Jain et al. demonstrated how the interface components of
the eco-feedback system influence energy consumption. The study correlated
users’ engagement with different interface tools to the reduction in energy
consumption (Jain, 2012). Gulbinas et al. experimented the effect of informing
commercial building occupants with their energy consumption through the
development of a modular socio-technical energy management system, BizWatts.
This system however, only provided the real-time data of plug-load appliances.
Nevertheless, the study concluded that providing the occupants with a network
level eco-feedback resulted in a statistically significant energy savings (Gulbinas,
2014). Another study was conducted by Emeakaroha et al. on students residential
halls, using IPTED (Integration of Persuasive Technology and Energy Delegate)
and real time energy feedback energy consumption reduction in 8 halls reached
37% amounting to savings of 1360.49 kWh, and 713.71 kg of CO2 compared to
the baseline (Emeakaroha, 2014).
In summary, University settings represented a convenient medium to
initiate energy performance tool prototype demonstration through actual energy
data monitoring and visualization. Data visualization provided comparison among
© ASCE
the different campus buildings, and as a result, it led to a more energy conscious
behavior from students, faculty members, staff, and most importantly better
understanding of the actual energy performance of building assets, thus improved
the operation process and promoted better ways to manage resources. The
emerging need by universities for clear data recently used a commercial data
visualization tool such as the Building Dashboard by Lucid Design Group
(Oakland, California), Energy Efficient Education Dashboard by Quality
Automation Graphics (Ankeny, Iowa), or conducted studies and developed their
own data analysis and visualization interface tools. The following (Table 1) is a
summary of selected studies relevant to this research:
Table 1. Summary of the utilization schemes in selected literature
(Emeakaroha, 2014)
(Domínguez, 2013)
(Yarbrough, 2015)
(Gulbinas, 2014)
(Malkawi, 2007)
(Dong, 2014)
(Jain, 2012)
Actual energy data ● ● ● ● ● ●
Simulated energy data ● ●
Peak energy analysis
● ●
and optimization
Calibrate energy models ●
Predicting future
● ●
consumption
Adjust the performance
of building systems/
● ● ● ●
Operations & facility
Management
Detect power faults and
●
deviations
Energy/economic
● ● ● ● ● ●
savings
Data
visualization/statistical ● ● ● ● ●
analysis
Online tool ● ●
Visualized data
availability for ● ● ●
occupants
Energy savings
possibility by
● ● ●
influencing occupants’
behavior
© ASCE
FR
RAMEWO
ORK
Thhis paper explains
e thee initial worrk on the developmen
d nt of a dataa visualization
daashboard thhat can bee of assistaance to booth campus operators and camppus
coommunity. The structuural framew work of the data visuallization boaard is dividded
innto three coomponents, the monitooring system componeent, the daata processing
coomponent, and a the dataa visualization component (see Fig gure 1).
Inn the first coomponent “Layer 1: Monitoring”,
M the data is collected frrom two maain
soources, the first sourcce is the university’s
u energy monitoring
m c
center, whiich
m
monitors thee electricityy consumptiion data onn the parceels level allong with tthe
ennergy used forf the chillled water inn the HVAC C systems. AsA for the water
w data, itt is
coollected man nually by thhe universitty facilities office whicch providess daily data of
w
water consum mption on thet campus and buildinng levels. In the secon nd componeent
“LLayer 2: Daata Processsing” the daata sorting and filterin ng and repoort generation
taakes place. InI the third componentt “Layer 3: Visualizatioon”, the datta is uploadded
onn the web online
o throuugh a data importer
i annd is compoosed of the interface anda
daata importerr.
Figu
ure 1. Fram
mework dessign of the data visuallization pro
ocess
Thhe raw dataa is presentted in user friendly deesign with comparisonn ability. TheT
tyype of data visualized
v o the website are simpple bar graphhs, pie charrts and gaugges
on
thhat contain energy annd water daata for each building or cluster of buildinngs
“pparcel”. Thee electrical energy connsumption isi presented d in kilowattt hour (kW
Wh)
annd the wateer is presennted in liteers. The tottal energy ofo the whoole campus is
caalculated ussing the colllected enerrgy data. Foor better beenchmarkin
ng, the enerrgy
usse intensity (EUI) is pllaced at thee main hommepage of th he website referenced
r by
yeear, where the
t EUI is thet total energy consum mption of a building divided
d by the
t
Wh/m2). Thhis kind of information
tootal area (kW i n gives the user
u the abiility to realiize
caampus energgy consumpption compaared to otheer campuses and buildinngs.
G
GRAPHICA
AL USER INTERFAC
I CE (GUI) DESIGN
D
Thhe user intterface for the system
m was devveloped usin ng the Sym mphony PH HP
framework, combined
c w the usee of the Boootstrap CSS
with S, jQuery JaavaScript, and
a
G
Google Charrts librariess. The userr-facing weebsite was divided intto three maain
seections: Ho
ome, Energgy Breakdoown, and Energy Map; M in adddition to tthe
© ASCE
addministrativve backendd which can c be useed to editt the diffeerent website

coonfigurationns and uplooad the dataa into the website.
w Thhe home page containss a
laarge digital meter
m displaying the overall camppus energy consumptio
c on per year, in
adddition to a color-codeed gauge too give the users
u a visuual represenntation of the
t
nuumber show wn in the meter.
m When n the users first view thet Home page,
p they are
a
shhown the daata for the current
c yearr, with the ability
a to seelect from previous
p yeaars
ussing a drop p-down conttrol. In adddition to thee meter andd gauge, thee Home paage
coontains a sets of charrts showingg the electtricity, watter, and co ooling enerrgy
coonsumptionn for each building
b in the selectedd year. Thee users havve a choice of
viiewing each of thosee charts, either in baar chart fo ormat with the averaage
coonsumptionn displayed as a line over
o the chhart; or as a pie chartt showing tthe
peercentage consumed by each of the t buildinggs for the currently
c selected enerrgy
tyype (see Figure 2, 3).
Figure
F 2. Bar w averagee consumption display
B chart with y
art display with perceentages

Figuree 3. Pie Cha
The homepage
h aalso contain
ns a control that is com
mmon acrosss most pagges
off the websitte, showingg the currentt month’s usage
u for a single buildding, for eaach
off the energyy types. Thhis data is displayed
d a a color-ccoded dial gauge,
as g withh a
seelector that can be usedd to switch between different builldings, or piick a building
att random. Every
E time a page is displayed to t the userr, a differennt building is
seelected at raandom and its
i data is displayed in this controll. The energ
gy breakdow wn
© ASCE
paage contain ns a similar dial gaugee control foor each buillding, with an additionnal
abbility to sellect which month
m to display
d fromm the data. This allowss the users to
coompare the historic ussage data forf this buiilding acrosss differentt months (ssee
Fiigure 4).
Figure 4. Data displaay using coolor-coded dial gauge
In adddition to thhe dial gauuge control, the energgy breakdow wn page allso
coontains a bar chart shhowing the energy connsumption of the seleected building
accross all thee months off a selectedd year, withh the option
n for the ussers to chooose
w
which year to t display. This
T gives the users ana overall view
v of the trends in the
t
buuildings’ ussage for eacch energy ty ype over thhe span of a year, and the ability to
coompare betw ween these trends
t acrosss different years (see Figure
F 5).
Figure 5. Trends in energy con

nsumption per
p month
Finallly, the energy map is designed

d to give users a visual ovverview of tthe
ennergy consuumption of the t entire caampus by displaying
d a map of thee campus wiith
eaach buildingg highlighteed in a colorr corresponnding to its energy
e conssumption (ssee
Fiigure 6). The
T map iss available in two diffferent moddes, one illustrating tthe
cuumulative energy
e usagge, and anotther one inddicating thee EUI, for each building
(ssee Figure 7).
7
© ASCE
Figurre 6. Cumu
ulative enerrgy consum
mption
Fiigure 7. Bu
uilding by building
b eneergy use in
ntensity (EU
UI)
The website
w inttegrated muultiple techhnologies too reach a high
h level of
fleexibility annd usability, clad in a modern useer interfacee design. Thhe Symphony
PH HP framew work was ussed as a baasis for the code, whicch offers a modular and a
m
maintainable way to devvelop the website
w efficciently and allow
a easy modificatioons
inn the future when needded. The in nterface waas implemennted using the Bootstrrap
CSS framew work, which offers flexxible and responsive usser interfacee capabilitiees,
cooupled withh the jQueery JavaScrript library,, which offfers easy accessa to tthe
poowerful funnctionality ofo JavaScriipt. The chaarts displayyed by the website weere
geenerated ussing the Gooogle Chartts (gChartss) chart librrary, which h enabled the
t
w
website to display
d cleaan-looking and dynam mic charts that allow the users to
innteract withh the data as
a they brow wse throughh the webssite and vieew the enerrgy
coonsumptionn at differentt times or fo
or different energy typees.
A
ASSESSME
ENT
A suurvey was conducted to evaluaate the impportance off the enerrgy

coonsumptionn data availaability to caampus users and the innfluence this availabiliity
caan have onn their behavior. Also o, this survvey was ussed to asseess the userrs’
peerception too the proposed visualizzation methhod versus the t tabular format. Sixxty
caampus userss took the survey
s incluuding staff, faculty meembers and students. TheT
suurvey was divided intto three seeparate secttions, the first f contained only thet
taabulated datta of the tottal energy consumptio
c on of each campus
c building throuugh
yeear 2014, in
n the secondd part the saame data waas given as a chart andd the third paart
w
was dedicated to the impportance of energy dataa availabilitty to all cammpus users.
Lookiing at the energy
e dataa tabulated and in a chart
c each at
a a time, tthe
suurvey takerss were requuested to staate their levvel of underrstanding off it on a scaale
from 1 to 10,, 1 being haardly and 100 very clearlly. Then another questiion was givven
© ASCE
too further asssess the levvel of underrstanding, the

t users were asked to
t state whiich
buuildings scoored an “aboove averagee” energy coonsumption.
Figure 8. Level off understan

nding of givven data
As seeen in (seee Figure 8 above), moost users foound the chhart easier to
innterpret, unllike the spreead sheet. The
T weighteed average level
l of undderstandingg is
9 for the charrt and 6 for the tabulateed data.
Figure 9. Chart and

d tabulated
d data repreesentation
The users weree asked “W Which buiildings scoored an abbove averaage

coonsumptionn?” and werre given tabbulated datta once and d chart dataa next. Maany
ussers (see Figure 9) failled to identtify the corrrect answerr, only 58.3% and 63.33%
piicked buildiings F and I respectiv vely. Buildinng C, being g just below
w the averaage
coonsumptionn was trickyy, thirteen reegarded it as
a above avverage whilee viewing tthe
taabulated datta, and only two users while
w viewiing the charrt. Using thee charted daata
(ssee Figure 10),
1 the useers picked buildings
b F and I as thhe building gs with highher
thhan average energy connsumption with
w a rate of o 96.5% an nd 90% resppectively. The
T
ovverall resullts showed that the charts are better
b repreesentation of
o the enerrgy
coonsumptionn data, most users could d fully undeerstand charts better thaan tables. Any
A
coonfusion thaat may occuur in cases like
l buildingg C, will bee eradicated d by the use of
innteractive chharts which give better insight on the
t data shoown.
© ASCE
F
Figure 10. Buildings
B haaving an ab
bove average energy consumptio
c on accordin
ng
to users’ choiices
CONCLUSIION
In thiis paper, wee have demmonstrated a new energgy visualizaation tool thhat
caan serve as educationaal tool for the
t campus communityy and potenntially impaact
beehavior by promoting better wayys to learn the energy y performannce issues on
caampus. Th he early feedback
fe frrom the campus
c ommunity support daata
co
viisualization. The futurre developm ment of thhe dashboarrd will be to includee a
caampus operaation decisiion support interface thhat integratee simulated building
b daata,
acctual and prredicted dataa per academmic year as shown in (see Figure 11). The errror
inn predictionn should be b minimizzed in the future as a result of o continuoous
reefinement off the simulaated modelss of campus buildings.
Figure 11. Propossed decision

n support in
nterface
R
REFERENC
CES
M
Malkawi, A.,, and Srinivvasan, R. (20007). “Enerrgy based deecision supp
port system
for faacilities mannagement: inntegration of
o data/web mining, kn nowledge baase
© ASCE
and thermal simulation.” Proceedings of Building Simulation, the

International Building Performance Simulation Association (IBPSA).
Jain, R. K., et al. (2012). “Assessing eco-feedback interface usage and design to
drive energy efficiency in buildings.” Energy and buildings, 48, 8-17.
Domínguez, M., et al. (2013). “Power monitoring system for university buildings:
Architecture and advanced analysis tools.” Energy and Buildings, 59, 152-
160.
Dong, B., et al. (2014). “Development and calibration of an online energy model
for campus buildings.” Energy and Buildings, 76, 316-327.
Emeakaroha, A., et al. (2014). “Integrating persuasive technology with energy
delegates for energy conservation and carbon emission reduction in a
university campus.” Energy, 76, 357-374.
Gulbinas, R., et al. (2014). “BizWatts: A modular socio-technical energy
management system for empowering commercial building occupants to
conserve energy.” Applied Energy, 136, 1076-1084.
Yarbrough, I., et al. (2015). “Visualizing building energy demand for building
peak energy analysis.” Energy and Buildings, 91, 10-15.
Building Dashboard. Lucid Design Group. Accessed 5 June 2015.
<http://www.luciddesigngroup.com/buildingdashboard/index.html>.
Energy Efficient Education Dashboard. Quality Automation Graphics. Accessed 5
June 2015. <http://www.qagraphics.com/energy-dashboard/>.
© ASCE
Using Relationship Mapping to Understand Sustainable Housing

Stakeholders’ Actions
S. Zedan1 and W. Miller2
1
Ph.D. Candidate, Science and Engineering Faculty, Queensland University
of Technology. GPO Box 2434, Brisbane 4001, Australia. E-mail:
sherif.lashen@hdr.qut.edu.au
2
Ph.D., Science and Engineering Faculty, Queensland University of
Technology, GPO Box 2434, Brisbane 4001, Australia. E-mail:
w2.miller@qut.edu.au
Abstract
The production of sustainable housing requires the cooperation of a variety of
participants with different goals, needs, levels of commitment and cultures. To
achieve mainstream net zero energy housing objectives, there is arguably a need for a
non-linear network of collaboration between all the stakeholders. In order to create
and improve such collaborative networks between stakeholders, we first need to map
stakeholders’ relationships, processes, and practices. This paper discusses compares
and contrasts maps of the sustainable housing production life-cycle in Australia,
developed from different perspectives. The paper highlights the strengths and
weaknesses of each visualisation, clarifying where gaps in connectivity exist within
existing industry networks. Understanding these gaps will help researchers and
practitioners identify how to improve the collaboration between participants in the
housing industry. This in turn may improve decision making across all stakeholder
groups, leading to mainstream implementation of sustainability into the housing
industry.
INTRODUCTION
Globally the key barriers inhibiting the residential construction industry’s
ability to implement sustainability requirements include technical, economic and
regulatory issues, the culture and practices of the industry, the lack of feedback loops
and poor levels of stakeholder engagement and communication (Holloway and
Bunker 2006; Williams and Dair 2006; Osmani and O'Reilly 2009). The realization
of efficient sustainable housing arguably requires the integration of functionality,
cultural sensitivity and local climatic conditions into a long-lived product (Larsson
2004), enabled by the implementation of sustainability practices throughout the
product life-cycle, from green field development and infrastructure provision,
through design, manufacture and construction, to ultimate deconstruction, re-cycling
or demolition (Romero et al. 2009; Miller and Buys 2012). This life cycle involves
© ASCE
the contribution of many stakeholders such as infrastructure providers, building

designers, constructors / contractors, product suppliers, policy makers and end-users /
occupants. Each stakeholder has different goals, needs, level of commitment and
culture that influence their participation in the construction sector and, as a result,
impact on the level of sustainability achieved by the end product.
Issues such as the lack of trust and communication among the participants,
the short term demands rather than the long term goals, the opportunistic behaviours,
and the lack of communication, contribute to producing low performance houses
(Tzortzatou 2007). Tzortzatou proposes that these problems call for the need to
follow a “partnering approach” that encompasses “Collaboration, open channels of
communication, and maximization of each participant’s recourse and expertise
through information and knowledge exchange” (Tzortzatou 2007). Other studies
show that firms that have trust-based, cooperative ties with their stakeholders are
more likely to succeed than firms that do not (Heugens, Bosch and Riel 2002).
Further literature suggests that a “relationship management” system that provides a
collaborative environment and a framework for all stakeholders to adapt their
decisions to project objectives is needed to expose hidden risks and maximize
sustainability outcomes (Cheung and Rowlinson 2011).
In order to optimize sustainable housing outcomes, there must be a non-linear
network of collaboration between all the stakeholders. This would enable essential
building information to flow between stakeholders, thus minimizing uncertainty
(Cheung and Rowlinson 2011) and influencing the decision making of all parties
towards more efficient and sustainable outcomes (Miller and Buys 2013). However,
the highly fragmented nature of the housing construction industry continues to
constrain development of these networks (Cheung and Rowlinson 2011; Epstein and
Widener 2011; Miller 2011), possibly due to the difficulty in developing a
collaborative tool that could be implemented in such a fragmented industry.
Rohracher suggests that socio-technical mapping can be used to identify the relevant
actors, technology and means of interactions and communication between those
actors, in order to understand and improve these networks (Rohracher 2001).
As an intermediate step between theory and practice, there is a need to
develop an understanding of the map that charts the players, their inter-relationships,
processes, knowledge, and goals. This paper will first describe a range of ‘mapping
frameworks’ that represent the sustainable housing production life-cycle in Australia
from different perspectives. These maps will be assessed to identify likely gaps in
connectivity between the relevant stakeholders. The paper will then propose a new
network map that closes these gaps, as a way to improve the collaboration and
information exchange between different stakeholders, towards greater sustainable
housing design and implementation in Australia
© ASCE
EVIEW OF
RE F AUSTRAL
LIAN HOU
USING NE
ETWORK MAPS
M
Thiis section prresents threee housing industry
i rellationship maps,
m from the perspecctive
of building innformation flows, connstruction industry reelationships and end-uuser
expperiences.
Maap1: Buildin ng Informaation Flowss and Stakeeholder Reelationshipss. This netw work
mapp (Figure 1)) depicts cuurrent comm mon practicees relating tot the creatiion and floww of
infoormation ab bout an inddividual dw welling from m site deveelopment thhrough to first
resaale to a sub T stages oof the housse production process are
bsequent puurchaser. The
represented onn the x axis (e.g. site developmeent, land saale, planning, construcction
etc)), whilst thee key playerrs involved in the proceesses are prresented in thet y axis. The
diaggram attem mpts to preesent, for eache stage of develop pment, which stakehoolder
creaates informmation abouut that speciific dwellinng, and to whom w this information is
passsed. This mapping
m shoows that noot all informmation abouut a dwellin ng is passedd on
throough the suupply chainn, arguablyy impactingg on decisio on making of subsequuent
stakkeholders. ItI identifies four key reelationships that impact on inform mation flowss: (i)
devveloper / in nfrastructuree provider (circular
( doot-dash linees); (ii) salees / valuatioon /
gulation / inndustry (squuare dotted lines); and (iv)
finaance (ellipse dotted linnes); (iii) reg
inittial purchaseer / subsequuent purchassers (ellipsee dot-dash liines).
Figure 1. Housing Informatio

I n Flows an
nd Relation
nships map..
© ASCE
Maap 2: Life Cycle

C Stagees Of New And Existting Sustaiinable Hou uses. A houssing
prooduction lifeecycle map (Figure 2) was initiateed by an Auustralian staate governm ment
authhority in 2014
2 and developed
d i a number of collaaborative workshops
in w w
with
houusing indusstry and asssociated suupply chainn stakehold ders. The purpose
p of the
mappping exerccise was too help gainn an undersstanding off the multip ple stakehoolder
perrspectives and
a networkk gaps in the t industryy. This maap representts the ongooing
prooduction lifee cycle of exxisting and new buildinngs, starting
g from manuufacturing until
u
demmolishing an nd reusing the
t materialls in the creeation of a new
n building g. It depictss the
keyy players innvolved in the creatio on and use of dwellin ngs, and thee collaboraation
betwween mem mbers of diifferent secctors of thee industry: resources,, research and
devvelopment, marketing,
m v
valuation, f
finance and real estate. Compared to the previious
mapp, this maap acknowlledges the role of reesearch as a key infl fluencer in the
prooduction of sustainable housing. The T map alsso considerrs the wholee life cycle of a
dweelling and th he managem ment of bothh new and existing
e hommes.
Figure 2.
2 Life Cycle Stages of New and Existing Su
ustainable Houses ma
ap.
© ASCE
Maap 3: A Su ustainable House

H As An
A
Useer-Centered d Integraated Systeem.
Thee third map m (Figuure 3) was w
devveloped by exammining a
and
evaaluating the experiences of end-useers
whoo were early adopters a of
conntemporary sustainablle homes in
subb-tropical Queensland.
Q This researrch
stroongly suggested that a sustainabble
houuse is an inntegrated system
s that is
cenntered on end-user goals and a
asppirations (th
he centre hexagon),
h a
and
incoorporates the intterconnectioons
betwween thesee goals andd aspirationss, Figure 3. 3 A sustaiinable housse integrateed

speecific buildding elemeents, desig gn system.
andd constructioon processees and practtices, and thhe urban co ontext (the six
s surroundding
hexxagons). Thhese interacctions take place
p withinn the contexxt of multiple supply chhain
ageents (the outter circle). Collectively
C y these com
mponents co ontribute to the productt - a
sustainable hoouse. This research
r furrther revealled that the success off the integraated
system is highhly reliant on good communicat
c tion betweeen stakehollders regardding
goaals, processees, and perfformance ou utcomes, annd on robustt decision suupport toolss.
GA
AP ANALY
YSIS OF AU
USTRALIA
AN HOUSIING NETW
WORKS
Whilst all three maps revveal
relaationships between the t Australlian
houusing indu ustry stakehholders, each
mapp portrayss differentt aspects of
houusing industtry collaborration (or lack
therreof). Mapaps 1 and 2 are indusstry
focuused, based on existting practicces.
Maap 4 (Figgure 4) representsr an
inteegrated verssion of thesse maps, whhere
the outer circlee representss the life-cyycle
phaases while the
t inner circle identifies
the key meembers in both maaps,
involved in each specifiic phase. This T
revised repressentation makes

m it eaasier to Fig
gure 4. Gaps in exissting practiices.
idenntify the many
m missinng direct chhannels (NNetwork den
nsity=0.6)
© ASCE
of information exchange between stakeholders within the construction life cycle (e.g.
between the real-estate market and designer or occupant and builder).
Three key gaps are revealed in these industry-focused maps: (i) where
information flows (or should flow); (ii) the direction of information flows; and (iii)
the (lack of) active involvement participation of end-users. Map 1 (Figure 1) revealed
that information created about a dwelling at different stages of its production is often
not passed on from one set of stakeholders to subsequent stakeholders. For example,
very little information is passed from the developer / infrastructure provider
relationship to other parties, while no information at all is passed from the sales /
valuation / finance relationship to the designer. Another example is the little
information passed to the subsequent owner, whose only source of information is the
sales agent who is likely not aware of most of the information. This lack of
information exchange leads to un-informed decision making or additional costs in the
re-creation of the information. One way to overcome this gap is to create a ‘building
file’ which holds all data sets that could be utilised by different members throughout
the building life-cycle. The building file concept is already being developed in
several EU-countries. It provides sets of information related to a particular dwelling,
such as description of the construction’s materials and elements, legal information,
condition report, quality reference, maintenance guide and energy label. These sets
of information support the buyers, occupants and policymakers in their decision
making process (Van der Bos and Meijer 2005). The building file concept could be
expanded to include sets of information that are distributed to all stake holders since
the inception of the project, using a variety of information generation, storage and
collaboration tools, such as building information model, energy simulation software,
building diagnostic tools, and post occupancy evaluation. Such a building file acts as
a storage hub that stores all the building data for future reference, and is updated
throughout the building life cycle to enhance transparency, so that the impact of
every decision is shown to all members.
Map 2 depicts many of the production stages as uni-directional (e.g. transition
from design to consultant to construction to marketing to new home buyer). A
bidirectional exchange of information would arguably enable each stage to provide
feedback loops to previous stages to enhance efficient and sustainable construction
outcomes. Robust building documentation systems and feedback loops, combined
with a ‘living lab’ approach, are needed in order to achieve the user-centered
aspirational view depicted in Map 3. The living lab involves the user in the creation,
exploration, experimentation and evaluation of innovations through active
involvement in all stages, including post-occupancy feedback (Romero, et al.,
2009).The living lab could be the link between maps 1, 2 and 3 in integrating
© ASCE
reseearch, user participatiion and thee on-going cycle of innformation flow betw
ween
reseearch and practice, occcupant and the
t housing industry.
TO
OWARDS A COLLAB
BORATIVE
E NETWO
ORK
Map 5 (Figure 5)) is a theorretical reprresentation of how gaaps in comm
mon
praactice miight be filled,
enhhancing thee bidirectional flow
of inform
mation between
stakkeholders during
d the different
houusing construction lifecycle
phaases througgh a netw work that
direectly conneect each staakeholder
to all the oth her stakehoolders. In
thiss map, eache key member
(inccluding e
energy e
efficiency
evaaluator that was added as a key
mem mber in the connstruction
phaase), has access to an
“infformation hub”. Thhis hub
conntains inforrmation colllected from m Figure 5. A Collaborativee Network

all the stakehoolders that are
a related tot (networrk density =1)=
the specific keey member and a can bettter inform the
t decisionns of that member.
m Tabble 1
gives examples on how thhe decisionss of each staakeholder arre affected byb other parrties
witthin the sam me networkk, and therrefore how better acccess to infoormation coould
impprove the paarticipationn of each staakeholder inn the housing network k. This netw
work
imppacts on different
d ouutcomes of the housinng industryy: strongerr networks are
exppected to leead to moree informed decisions which
w in turrn should lower risks and
resuult in higherr performannce sustainaable homes.
Tablee 1: Examp
ples of Collaaborative Communica
C ation Netw
works (Map 5)
line Memberrs Sample Relationsh hip Descrip
ptions
1 Occupannt/policy Policies are made for
f the safeety and well-being of the
t
and reguulation occupan nt.
Occupannt awareneess of thee regulation n assists his
h
choice.
2 Occupannt/developerr Developper makes investments
i s based on the needs of
targetedd occupant.
3 Occupannt/designer Occupannts’ needs and behaviours influeence designner
(user-centred designn).
d wellbeing of
The design influencces the behaviours and
the occuupants.
© ASCE
4 Occupant/builder Quality construction (insulation, airtightness)

influences occupant’s behaviour and enhance
wellbeing. Occupant feedback can influence quality
construction.
5 Occupant/supplier Quality of materials can influence occupants’ decisions
and health.
6 Occupant/simulator Occupancy (number and behaviour) impacts on
simulation results.
Feedback from occupants can influence simulation
development.
7 Occupant/real- Occupants influence as well as are influenced by, the
estate market market value.
Real estate agents can promote sustainable housing
benefits to occupants
8 Designer/developer Designer incorporates aspirations of the investor into
(investor) his designs, and integrates sustainability in a way that
does not compromise the goals of the investor
9 Designer/ regulator Regulations stipulate minimum design and
construction standards.
Designers can exceed regulatory requirements for best
practice.
10 Designer/real-estate Designers and real estate agents can respond to, and
market create, market demand for sustainable housing.
11 Designer/simulator Ongoing modification in the design based on the
simulation results could achieve optimal sustainability
performance at least cost.
12 Designer/supplier Designer awareness of the lifecycle impacts of
construction products and locally available sustainable
materials.
13 Designer/builder Ongoing bidirectional relationship to adapt the design
to circumstances without compromising performance
quality and ensuring construction is as per design.
14 Builder/developer Developer goals should not be altered during
construction.
15 Builder/regulator Construction is bound by the regulations. Regulators
should ensure compliance.
16 Builder/real-estate Quality construction leads to higher market value.
Valuation strategies need to incorporate sustainability
characteristics.
17 Builder/simulator Simulator modifies simulation based on ‘as
constructed’ information, verifies existing situation is
similar to design, and gives evaluation feedback to
builder
18 Builder/supplier Quality of supplies is assessed by builder.
Supplier assists in the quality implementation of his
products into construction.
© ASCE
19 Supplier/regulator Regulations regarding the use of materials, production

and transportation.
20 Supplier/real-estate Knowing products and materials quality will assist in
the valuation process.
21 Supplier/ simulator Knowledge of material impact on energy performance.
22 Simulator/developer Simulation can influence initial decisions of the
developer.
23 Simulator/regulator Construction to be checked for compliance with
simulation’s energy rating.
24 Simulator/real Simulation results (e.g. energy performance
estate market certificates) can be used as a sales promotion document
by real estate agents.
25 Real-estate Consideration of mechanisms and tools to value the
/developer project in the current market and future markets (over
the lifecycle of the dwellings / estate).
26 Real- Agents are bound to regulations when promoting or
estate/regulator contracting a property.
27 Developer/supplier Developer should invest in using and developing
sustainable materials.
28 Developer/Regulato Developer should follow regulation. Regulation should
r encourage investments.
CONCLUSION
The paper examined two maps that represented existing common practice and
stakeholder relationships in the Australian housing industry and a third map, centered
on the end-users’ perspectives. From the analysis of these maps it became evident
that there are gaps in the information network that connect the stakeholders to each
other throughout the housing production life-cycle. These gaps could impact the
sustainable housing outcomes. The paper then represented a map which reflects the
need to treat sustainable housing as an integrated system, where collaborative
processes are made to reach solutions that maximize sustainable housing outcomes.
A further development of this paper could be the integration of maps three and five,
where the information from each member’s related “information hub” in Map 5 is
used to make informed decisions that contribute to the fulfilment of a centeral goal
which is the sustainable house aspirations of the occupants.
REFERENCES
Cheung, Yan Ki Fiona and Steve Rowlinson. 2011. "Supply chain sustainability: a
relationship management approach." International journal of managing projects in
business 4 (3): 480-497.
Epstein, Marc J. and Sally K. Widener. 2011. "Facilitating sustainable development
decisions: measuring stakeholder reactions." Business strategy and the environment
20 (2): 107-123.
© ASCE
Heugens, P. P. M. A. R., van den F. A. J. Bosch and van C. B. M. Riel. 2002. "Stakeholder
integration." Business & Society: a journal of interdisciplinary exploration 41 (1):
36.
Holloway, Darren and Raymond Bunker. 2006. "Practice Reviews: Planning, Housing and
Energy Use: A Review." Urban Policy and Research 24 (1): 115-126.
Larsson, Nils. 2004. "The Integrated Design Process." Accessed 14 December 2010.
http://www.iisbe.org/down/gbc2005/Other_presentations/IDP_overview.pdf.
Miller, Wendy. 2011. "Greening Australia's residential built environment: Part A":
SBENRC.
Miller, Wendy and Laurie Buys. 2012. "Positive Energy Homes: Impacts on, and
Implications for, Ecologically Sustainable Urban Design." Urban Design
International 17 (1): 45-61.
Miller, Wendy and Laurie Buys. 2013. "Factors influencing sustainability outcomes of
housing in subtropical Australia." Smart and Sustainable Built Environment 2 (1):
60.
Osmani, Mohamed and Alistair O'Reilly. 2009. "Feasibility of zero carbon homes in
England by 2016: A house builder's perspective." Building and Environment 4 (9):
1917-1924.
Rohracher, Harald. 2001. "Managing the Technological Transition to Sustainable
Construction of Buildings: A Socio-Technical Perspective." Technology Analysis &
Strategic Management 13 (1): 137-150. Accessed 2015/05/22.
Romero, David, Myrna Flores, Carlos Vallejo and Arturo Molina. 2009. "Towards a Novel
Living Lab Model for Sustainable Innovation in the Construction Industry." In 15th
International Conference on Concurrent Enterprising (ICE 2009), Leiden, The
Netherlands, 22-24 June 2009, edited.
Tzortzatou, Eleni Penelope. 2007. "Culture impact in construction supply chain
management." Dissertation/Thesis, ProQuest, UMI Dissertations Publishing.
Williams, Kate and Carol Dair. 2006. "What is stopping sustainable building in England?
Barriers experience by stakeholders in delivering sustainable developments."
Sustainable Development 15 (3): 135-147.
© ASCE
The Weatherization Assistance Program:

Social Policy or Energy Policy and Why It Matters
J. N. Terman1
1
George Mason University, School of Policy, Government and International Affairs,
4400 University Dr., Robinson B, 3F4, Fairfax, VA 22030. E-mail:
Jterman@gmu.edu
Abstract
Low-income populations spend a larger proportion of their income on energy

usage and tend to live in housing stocks that consume more energy. Thus, energy
efficiency is cross-cutting goal, connected to both social and energy-related concerns.
One case in point is the Weatherization Assistance Program (WAP): a federal
program that assists low-income families through education, energy audits and
retrofits. While some states have exerted considerable effort to achieve program
goals, others have had lackluster results. The study results here suggest that ambitious
program – measured as the proportion of home weatherizations planned that were
completed – is influenced by whether states view the program as a social policy-
oriented program.
INTRODUCTION
Countries all over the world have been increasingly sensitive to their energy
usage and costs. Building energy costs account for 41% of total energy consumption
costs (DOE, 2010). One way to reduce these costs is by engaging in energy retrofits
and weatherization to ensure building stocks are using energy as efficiently as
possible (Dietz, et al., 2009). This is particularly important for low-income
households, who dedicate a larger proportion of their incomes to energy costs
(ORNL, 2010). Thus, energy efficient housing is not only an energy usage issue of
energy efficiency, but it is also a social policy concern. The Weatherization
Assistance Program (WAP) demonstrates this duality of purpose. The WAP is a
federally funded, state implemented program designed to reduce the energy
consumption costs of low-income households through energy audits and retrofits.
While seemingly straightforward, some states view the program as a social services
program aimed at the welfare of the target population. Other states view it as a
vehicle for primarily energy efficiency (EE) or economic development. This study
tests whether these program framing choices influence WAP performance.
© ASCE
ENERGY EFFICIENCY HOUSING AS A SOCIAL AND ENERGY POLICY

ISSUE
The benefits accrued by the behavioral and economic changes of
weatherization are not evenly distributed across the population. American families in
the lowest income brackets are increasingly paying a larger proportion of their
income in energy costs. Households with an average yearly income of between
$10,000 and $30,000 pay 24% of this income on energy consumption (America’s
Power, 2012). Since energy costs are fixed across income groups, energy
consumption costs can be very regressive (Eisenberg, 2010). Thus, EE has become a
social issue. The WAP is designed to help low income families reduce their energy
costs through energy auditing, retrofits and weatherization. Federal funds are
funneled from the Department of Energy (DOE) to state agencies where program
offices are housed. These state agencies are responsible for providing support and
training local service providers – government agencies, private organizations, and
nonprofit organizations – that conduct energy retrofits in the service areas defined by
the state.
In 2009, through the American Recovery and Reinvestment Act (ARRA), the
WAP received a $5 billion infusion of federal dollars to provide increased
weatherization services to be spent by March 2012. In return, states set
weatherization performance goals that were to be achieved during that fixed period of
time. In order to meet this increased demand for weatherization, states also received
an increase in training and technical assistance dollars that was intended to help states
train hard-to-employ populations and meet their performance goals. Nonetheless,
there was enormous variation in goal achievement. What I suggest is that, through
agency assignment, staff assignment and ancillary program support, states can frame
the WAP as a social policy program – as opposed to program focused on energy. This
framing influences the prioritization of program goals in addition to the
implementation infrastructure that facilitates the achievement of those goals.
AGENCY ASSIGNMENT AND RESOURCE ALLOCATION
This study takes inspiration from the literature on agency assignment (Krause,
Feiock and Hawkins, 2015) and the organization of attention (May, Workman and
Jones, 2008; Terman, 2014), which suggests that, in making choices about resource
investments and bureaucratic structure, governments are sending messages about how
they frame and prioritize public programs. One way to think about agency assignment
and structure is to consider the substantive location of agency responsibilities. As
large organizations with multiple goals, agencies have a significant array of policy
responsibilities. However, with limited information processing capabilities (May,
Sapotichne, and Workman, 2009), agency structures will favor some agency
responsibilities over others. For example, May, Workman and Jones (2008) suggest
that, when multiple responsibilities were moved into the Department of Homeland
Security (DHS), the agency was able to respond to terrorism concerns but had policy
failures in other areas, such as emergency management (e.g. Hurricane Katrina). The
information processing pathways of the organization were unable to handle the
diversity of policy responsibilities. The implication is that the location of agency
responsibilities within a substantive policy area influences a government’s ability to
carry out those responsibilities.
© ASCE
Governments make choices about how they frame their policy responsibilities
(Krause, Feiock, and Hawkins, 2014). These choices have substantive consequences
in that an important causal mechanism for program performance is the
implementation structures that are shaped by these choices. For example, agencies
that have a substantive focus on economic development will be more likely to
prioritize the programs that have heightened results in terms of economic
development.
Applying these ideas to the WAP, the question becomes whether state
governments frame the program in terms of social policy? This is important because,
as an EE-oriented policy, the return on investment (ROI) for the program is fairly
low. While programs within DOE’s Energy Efficiency and Renewable Energy
(EERE) Office portfolio have historically had a 20 to 1 benefit to cost ratio, WAP has
a benefit to cost ratio of approximately 1.53 to 1 (DOE, 2009a). In the EE policy
world, this ROI helps determine program and policy prioritization. Programs with
lower ROIs become less of a priority. Thus, agencies focused on energy issues may
be less likely to ambitiously implement low-ROI programs such as the WAP.
ROI in social programs can be defined more broadly because the program
intervention has goals that are often more focused on societal value sets. For
example, the American Public Human Services Association (APHS, 2013) has
developed the idea of a social return on investment (SROI). SROI is calculated as a
blended value of the financial and social benefits of particular policies and programs.
These broader social goals are more likely to be taken into consideration and valued
by agencies focused on social services and administering to vulnerable populations –
in comparison to agencies focused on energy or economic development. Thus, the
expectation is that, when the WAP is nested in agencies that have a social service
orientation, program performance will be higher (H1).
Agency assignment is not the only way that states can frame their policies;
they can also engage in staff assignment. They may dedicate staff with particular
backgrounds to administer programs. For example, governments may choose WAP
staff with expertise in energy efficient housing stocks that would see the program in a
different light than those with a background in social services or anti-poverty. In the
WAP, governments can engage in staff assignment by having the program directed
by administrators involved in the delivery of other social services within those
agencies. One way that they do this is by leveraging their administrators from the
Community Services and Block Grant Program (CSBG). CSBG is touted as one of
the only federal programs that is exclusively dedicated to the reduction of poverty
(NASCP, 2012). The professional network of administrators associated with CSBG
programs is robust and well-coordinated with active lobbying and training arms
(NASCP, 2009).
States have the opportunity to use CSBG trained administrators to implement
the WAP. This is the case even if these administrators are not in social service-
oriented agencies. When CSBG administrators direct the WAP, these individuals will
be more focused on the social service aspect of the program – rather than EE –
thereby heightening program performance. The expectation is that, when the WAP is
directed by CSBG administrators, program performance will be higher (H2).
© ASCE
The larger causal mechanism behind agency and staff assignment is the
substantive prioritization of particular policy goals (and the programs associated with
those policy goals). One indicator of this prioritization is the ancillary support
associated with the program. As a social program, an overarching goal of the WAP is
to not only reduce energy costs but to also train the hard-to-employ with
weatherization skills that are then utilized to provide program weatherization services
(OWIP, 2009). To complete this training, states were mandated to spend 20% (up
from 10%) of their ARRA-allocated WAP dollars on training and technical assistance
(T&TA). However, this training comes in different forms and strengths.
T&TA centers are considered one of the most vigorous training methods
available for building the weatherization workforce of the hard-to-employ (OWIP,
2009). They are specifically intended to train the hard-to-employ in weatherization
centers that consist of classrooms and laboratory space set up for long-term training.
DOE documentation suggests that training centers are the most robust method
through which states can meet their weatherization goals because they are more likely
to reach the hard-to-employ. Attendance can be mandated and training recipients can
be held accountable, tracked and supported in terms of job placement (OWIP, 2009).
While T&TA centers are an extremely robust way to train the hard-to-employ,
there are other less permanent educational methods for training. States may also
choose to invest T&TA dollars in training WAP enrollees, as opposed to curriculum
and website development, facility preparation and overhead. While the latter are
important investments, they do not go directly to training the hard-to-employ. These
investments are indicative of the broader view that the WAP is a social-services
oriented program with a broader scope to reduce the sources of poverty – not only
high energy costs, but also unemployment and lack of job skills. As such, the
expectation is that, when states choose to invest T&TA dollars on activities that focus
on training the hard-to-employ, as opposed to support services, program performance
will be higher (H3).
RESEARCH DESIGN
The dependent variable in the study comes from DOE and State Energy
Program (SEP) offices, which are responsible for maintaining records on the WAP’s
performance goals (and achievement of those goals). As a condition of receiving
ARRA dollars, states had to set performance goals – the number of weatherizations
planned – with their service providers. Performance goal achievement is measured
here as the proportion of weatherizations planned which were completed.
The independent variables of interest include (1) whether the WAP is
administered through an agency that is centered on social services, (2) whether the
WAP is directed by CSBG administrators, (3) whether the state has chosen to invest
in one or more T&TA centers and (4) the proportion of T&TA dollars that are
allocated to direct training of WAP enrollees. The former was operationalized by
examining the state agencies administering the WAP. There are three types of
agencies that administer the WAP: social services-oriented agencies, economic-
development focused agencies and energy focused agencies. Since agency
assignment varies so much across states, observations were coded by whether the
© ASCE
WAP was part of an agency whose target population is persons in poverty. If the
WAP was outside of the primary agency assigned to service vulnerable populations,
then the agency was not coded as a social services-oriented. The data for the second
independent variable of interest was obtained from the 2009 CSBG annual report
while the source for the third and fourth variables of interest were obtained from
Department of Energy performance reports for T&TA resource use.
Service area characteristics and state partisanship, size, resource availability
and professionalism were controlled for in the analysis. State characteristics were
particularly important because of the potential for them to influence investment a
T&TA center and propensity for goal achievement. Table 1 includes a complete
listing of the control variables, operationalizations and descriptive statistics.
The unit of analysis in this study is the WAP service area; service areas are
nested within states. Thus, a hierarchical linear model is employed. After
considerable experimentation with model specification, the decision was made to
estimate a model with a random intercept and random slopes for T&TA investments
and agency assignment. When modeled as random slopes, the random effects for staff
assignment and the proportion of WAP training dollars allocated to direct training
were not statistically different from zero and were removed from the model
specification.
RESULTS
As Table 1 shows, the average number of weatherizations planned that were

completed was 1.329. There are 630 service areas that reside in states with at least
one T&TA center. Nine states administer the WAP through energy-focused agencies;
thirteen administer it through social-services focused agencies; and 28 administer it
through economic development-oriented agencies. The model results are depicted in
Table 2. The model yielded good statistical fit. The Wald test is statistically
significant at p < 0.001. The results indicate that the effects of agency assignment in a
social services-oriented agency and agency assignment in an economic development-
oriented agency, in addition to investment in at least one T&TA center are
statistically significant and vary across states. However, the variation across state
slopes for WAP assignment to a social services-oriented agency is quite small.
In terms of H1, when the WAP is administered through a social services
agency, state performance goal achievement increases by an average proportion of
0.351 in comparison to when the WAP is administered through an energy-oriented
agency (controlling for other variables in the model). As the random effect suggests,
this relationship differs minimally across states. However, in a somewhat surprising
result, when the WAP is administered through an economic development-oriented
agency, goal achievement increases by an average proportion of 0.336. This is, on
average more than 0.15 below the effect for a social services-oriented agency.
Furthermore, this effect varies by roughly 0.25 across states. This suggests that there
is more variability in performance goal achievement for states that have the WAP
administered through economic development-oriented agencies. This variability
introduces the possibility that the effect that economic development-oriented agencies
have on goal achievement is greater than that of social service-oriented agencies.
© ASCE
However, in comparison to energy oriented agencies, the effect is considerable for

both.
The model yielded no statistically significant results for H2, whether or not
the WAP was directed by CSBG administrators. In terms of one of the variables
testing H3, when states invest in a T&TA center, state performance goal achievement
increases by an average of 0.18 (controlling for other variables in the model). The
second variable testing H3, amount of T&TA money spent per WAP enrollee in job
training was not statistically significant in the model.
Table 1: Codebook and Measurement

Measurement Mean Min Max
(SD)
Performance goal Total weatherizations completed 1.329 0 7.214
achievement Total weatherizations planned (0.540)
Source: DOE/WAP1
Social Services Orientation
Agency assignment Social-services-oriented agency 0.2 0 1
Source: Agency mission statements (0.4)
Economic development-oriented 0.663 0 1
agency (0.473)
Energy-oriented agency 0.137 0 1
(reference category) (0.344)
Staff assignment WAP directed by CSBG administrators 0.525 0 1
Source: 2009 CSBG Annual Report (0.499)
T&TA center > One T&TA Center 0.654 0 1
Source: DOE/WAP (0.476)
WAP job training Proportion of individuals enrolled in a 0.273 0 1
activities T&TA activity that received job (0.444)
training
Service-Area Characteristics
% Eligible % WAP eligible population 15.28 0 1
Source: Census 2009 (5.042)
Housing density Housing units per square mile 98.512 0.5 483.2
(99.112)
WAP allocation ARRA-funded WAP allocation 4488501 47756 1.00e+
Source: DOE/WAP (5242429) 08
Service provider For-profit provider 0.021 0 1
type (0.143)
Government provider 0.174 0 1
(0.380)
Nonprofit provider 0.804 0 1
(reference category) (0.397)
State Characteristics
Gubernatorial Democratic governor in 2009 0.632 0 1
partisanship Source: BOS2 2009 (0.483)
Legislative % Democrats in State legislature 49.847 24.67 70.55
partisanship (8.857)
Population density Residents per square mile 230.651 15853 29766
© ASCE
Per capita income Population in the state 21222.95 15853 28766

Unemployment % Unemployed 8.978 4.1 13.4
3
Source: BLS (1.752)
Professionalism Government Performance Project 6.105 2 9
score (1.442)
Source: PEW Charitable Trusts
1
All data originating from DOE and WAP offices are in terms of ARRA funding for the
implementation period (2009-2012).
2
Book of States
3
Bureau of Labor Statistics
DISCUSSION AND CONCLUSION
The study has yielded interesting findings. First, the differing effects between
the types of agencies administering the WAP are considerable. The essential
difference between social services and economic development-oriented agencies and
those focused on energy is that the former types deal with social problems. This is
fundamentally different than energy-oriented agencies, which often focus on natural
resource issues. The infrastructure and skill sets required to deal with the issue of
poverty are fundamentally different, and will lead to performance differences.
Table 2: Performance Goal Achievement
β SE
Social Services Orientation
Social services-oriented agency 0.339*** 0.096
Economic development-oriented agency 0.325*** 0.093
WAP directed by CSBG administrators 0.091 0.098
T&TA center 0.184** 0.086
WAP job training activities 0.034 0.1
Service-Area Characteristics
% Eligible -0.01*** 0.004
Housing density 0.007 0.007
WAP allocation
For-profit provider -0.213** 0.109
Government provider 0.123*** 0.42
State Characteristics
Democratic governor 0.067 0.082
% Legislative Democrats -0.009* 0.005
Population density -0.003 0.003
Per capita income < 0.00 < 0.00
% Unemployment -0.012 0.02
Government Performance Project score -.048* 0.025
Random Effects
Social services-oriented agency 0.001 < 0.00
Economic development oriented agency 0.261 0.092
T&TA center 0.23 0.081
Constant 1.03 0.444
Service area effects 0.456 0.011
Service areas/State Service areas = 962 Wald χ2 (16) = 65.23***
© ASCE
Min = 3 States = 50 Log Likelihood = -649.033

Max = 74
Avg. = 19.2
Note: *** p<=.01, ** p<=.05, * p<=.1
The results of having at least one T&TA center are more intuitive. With
investment in stand-alone T&TA centers, there may a greater commitment to the
broader program purpose of poverty reduction. However, this result may be tenuous
because there may be other reasons for investing in a T&TA center in a given state.
For example, some states may choose not to invest in a T&TA center because a
neighboring state has one (i.e. Texas and Arizona do not have a T&TA center but
New Mexico does). Although, T&TA centers are intended for training the hard-to-
employ from inside the state – unless they are regional centers – states may send
individuals across state lines for training.
The results here are relatively persuasive. Service areas in states whose WAPs
are served by social services or economic-development-oriented agencies, on
average, have one-third higher goal achievement. Services areas in states with T&TA
centers have nearly one-fifth higher performance goal achievement. Nonetheless,
these results highlight some important challenges to the development and
implementation of energy efficiency and sustainability programs. These programs
have environmentally collective benefits but depend on a concentrated population
that is traditionally difficult to administer to. Appropriate program design requires
expertise in both energy policy and social services. The development of
weatherization techniques and training – by experts in the energy and building
industry – is essential to the services provided by the program. However, program
services and overall success cannot occur without considerable knowledge of the
target population, to whom and by whom services are rendered. Perhaps, if energy-
oriented agencies are involved in higher ROI energy programs, it makes sense for
social services-oriented agencies to have the responsibility for program
implementation so that the former can use their expertise to focus on programs with
broader target populations and higher return. However, in the context of federally
funded, locally implemented programs, lower level governments make these choices,
which is what accounts for this variation.
Future research can build on these ideas and address some of the limitations in
this study. The operationalizations used here are somewhat compelling in terms of
viewing the WAP as a social program, however, they need to be broadened and the
causal mechanism behind them needs to be better understood. For example, what is it
about the agency program implementation structures that affect performance? Is this
an issue of heightened staff motivation to achieve program goals? With a low ROI in
energy-related program such as the WAP, some agencies are simply not incentivized
to dedicate significant resources toward its implementation.
REFERENCES
American Public Human Services Association. (2013). Social Return on

Investment.APHS Innovation Center: Washington, DC.
© ASCE
Department of Energy (2009a). Congressional Budget Request, Volume 3.

Government Printing Office: Washington, DC.
Dietz, T., Gardner, G. T., Gilligan, J., Stern, P. C., & Vandenbergh, M. P. (2009).
Household actions can provide a behavioral wedge to rapidly reduce US
carbon emissions. Proceedings of the National Academy of Sciences, 106(44),
18452-18456.
Eisenberg, J. (2010). Weatherization Assistance Program Technical Memorandum
Background Data and Statistics. Oak Ridge National Laboratory: Oak Ridge,
TN.
Energy Information Administration. (2014). Frequently Asked Questions,
http://www.eia.gov/tools/faqs/faq.cfm?id=86&t=1.
Krause, R. M., Feiock, R. C., & Hawkins, C. V. (2014). The administrative
organization of sustainability within local government. Journal of Public
Administration Research and Theory, muu032.
Illinois Association of Community Action Agencies. 2010. 2010 Illinois Weather
Assistance Program Report. Department of Commerce and Economic
Opportunity: Springfield IL.
May, P. J., Sapotichne, J., & Workman, S. (2009). Widespread policy disruption:
Terrorism, public risks, and homeland security. Policy Studies Journal, 37(2),
171-194.
May, P. J., Workman, S., & Jones, B. D. (2008). Organizing attention: Responses of
The bureaucracy to agenda disruption. Journal of Public Administration
Research and Theory, 18(4), 517-541.
McKenzie-Mohr, D. (2013). Fostering sustainable behavior: An introduction to
community-based social marketing. New society publishers.
National Association for State Community Services Programs. (2012).
Weatherization Assistance Program Annual Funding Survey. NASCSP:
Washington, D.C.
National Association for State Community Services Programs. (2009).
Weatherization Assistance Program Annual Funding Survey. NASCSP:
Washington, D.C.
Patterson, M. G. (1996). What is energy efficiency?: Concepts, indicators and
methodological issues. Energy policy, 24(5), 377-390.
Office of Weatherization and Intergovernmental Program. (2009). National
Weatherization Training and Technical Assistance Plan. Washington, D.C.:
Department of Energy.
Terman, J. (2014). A State-Level Examination of Bureaucratic Policymaking The
Internal Organization of Attention. The American Review of Public
Administration, 0275074014529840.
© ASCE
Towards Multi-Objective Optimization for Sustainable Buildings with Both

Quantifiable and Non-Quantifiable Design Objectives
W. Yan1; M. Rahmani Asl2; Z. Su3; and J. Altabtabai4

1
Department of Architecture, Texas A&M University, 3137 TAMU, College Station, TX
77843. E-mail: wyan@tamu.edu
2
77843. E-mail: mrahmaniasl@gmail.com
3
77843. E-mail: zhouzhousu@gmail.com
4
77843. E-mail: jtabtab@me.com
Abstract
Architectural design can be seen as a multi-objective optimization process,

aimed at finding optimal solutions for multiple, often conflicting objectives. Despite
the significant advancements of multi-objective optimization methods and building
modeling technologies, multi-objective optimization has not been enabled for
building design in practice. The reasons include: 1) the complexity of fitness
evaluation for quantifiable objectives, such as structural, economical, and sustainable
performance objectives, in which a large number of building design decision
variables are discrete, nonlinear, and stochastic; and 2) the missing of non-
quantifiable objectives of overriding significance to architects, e.g. aesthetic, social,
and cultural objectives in optimization. The research objective of this research is to
discover how to develop a parametric BIM-based, multi-objective optimization
framework with both quantifiable and non-quantifiable objectives, and how it can
search for Pareto Optimal design solutions with diverse non-quantifiable objective
measures to assist architects’ decision-making.
INTRODUCTION
Sustainability with efficient energy use and minimal environmental impact

has become a major objective in building design, adding to other traditional
objectives of architecture: aesthetic, structural, functional, economical, social,
cultural, historical, behavioral, etc. Despite the significant advancements of multi-
objective optimization methods and building modeling technologies, optimization has
not been enabled for building design in practice. The reasons include: 1) the
complexity of fitness evaluation for quantifiable objectives, in which a large number
of building performance decision variables are discrete, nonlinear, and stochastic; and
2) the missing of essential non-quantifiable objectives in optimization, e.g. the
objective of overriding significance to architects - aesthetics. The goal of this
research is to advance multi-objective optimization for creative and sustainable
© ASCE
building design. To achieve the goal, we propose an innovative Architectural Multi-

Objective Optimization (ArchMOO) framework using Evolutionary Algorithms and
Parametric Building Information Modeling (BIM).
BACKGROUND AND PROBLEM STATEMENT
According to the earliest surviving written work on the subject of architecture

De architectura by the Roman architect Vitruvius in the 1st century BC (Pollio 1914),
a good building should achieve objectives in the three principles of firmitas, utilitas,
venustas, or in English: durability, utility, and beauty. More broadly, architecture
should promote aesthetic, social, environmental, traditional, gender-based, economic,
mathematical and scientific, and the novel design values (Holm 2006).
In general there are two methods to solve multi-objective optimization
problems: 1) Convert multiple objectives into a single objective by using a weighted
sum of the objective functions with often arbitrary weights; and 2) Search for the
Pareto Optimal set, which is more widely accepted for practical MOO, as normally
there exist a set of optimal solutions (design options) instead of a single solution for
MOO. A change to a design option’s parameter values that makes at least one
objective better off without making any other objectives worse off is called a Pareto
Improvement. A design option is defined as "Pareto Optimal" when no further Pareto
Improvements can be made (Radford & Gero 1987). There can be multiple design
options, each of which meets the following condition: any improvement of one
objective, e.g. decreasing the use of operating energy, will degrade at least one other
objective, e.g. increasing the life-cycle cost of the building. These design options are
the Pareto Optimal Set that we seek to find using computational methods.
Multi-Objective Genetic Algorithm (MOGA) has become a major MOO
method in many areas of science and engineering. The common characteristics of the
problems that can be resolved by MOGA are a set of quantifiable objectives, a set of
constraints, and a process of obtaining trade-off information between the stated
objectives (Coello et al. 2007). Based on the Pareto Optimal solution set found by
MOGA, designers can analyze the solutions and make a design decision as to further
develop which of the solutions into a final design.
However, despite the rapid advance of MOGA algorithms (Coello et al. 2007)
and the extensive applications of MOGA in many fields of science and engineering,
the application of MOGA in architectural design practice is quite limited. The reasons
include: 1) the complexity of fitness (building performance) evaluation for
quantifiable objectives, and 2) the missing of essential non-quantifiable objectives in
optimization.
Complexity of fitness evaluation for quantifiable objectives. An important

component of MOO is fitness evaluation. Among architectural performance
evaluations, energy performance evaluation is highly complex. This complexity
comes from the large number of interrelated parameters such as building geometry,
space layout and functions, materials, sites, climatological data, user behaviors, and
many engineering calculations based on differential algebraic equations (DAEs) that
© ASCE
are used to evaluate the energy performance of buildings (ANSI/ASHRAE 2007;

Torcellini et al. 2006; Bazjanac 2001).
Missing essential non-quantifiable objectives in MOGA. Non-quantifiable

objectives that architects value highly are essential design objectives (Kellert et al.
2011; Lera 1980). Among these design values, aesthetics and self-expression are the
artistic aspects characterized by one’s inner spiritual self and creative imagination,
inner resources and intuition, which should be the base used when designing (Holm
2006). Aesthetics are one of the strongest non-quantifiable components of
architectural design (LaHood & Brink 2009). For many architects, aesthetics are of
overriding significance (Ruskin 1989). In architecture, however, only limited studies
have experimented MOGA and Pareto Optimality for building design with
quantifiable objectives (e.g. Caldas 2006, Welle et al. 2012, Gerber et al. 2013,
Bradner and Davis 2013).
In light of the complexity of evaluation for quantifiable objectives (e.g.
energy) and non-quantifiable objectives (e.g. aesthetics), there is a significant need to
research and develop an advanced MOGA for architectural design. We know of no
other studies that have included non-quantifiable objectives, e.g. aesthetics, in
MOGA for architecture. Therefore we propose an Architectural Multi-Objective
Optimization framework (ArchMOO), which will incorporate non-quantifiable
objective comparison algorithms (e.g. shape comparison algorithms) into MOGA in
order to increase the diversity of those objectives (e.g. the building forms) in the
Pareto Optimal solutions that will significantly help designers participate in the
optimization process, and in the meantime, drastically reduce the computation needed
by MOGA.
METHODOLOGY
Our research methodology consists of the following aspects: investigate,

prototype, and evaluate the proposed human-computer cooperative, theoretical
ArchMOO framework, which consists of a Genetic Algorithm for Pareto
Optimization emphasizing diversity of non-quantifiable objective measures
(architectural aesthetics concerning building forms as a case study). Below we
describe the elements of ArchMOO and the rationales behind developing these
elements, technical problems associated with each of them, and methods to make the
elements particularly well suited for ArchMOO to enable multi-objective
optimization for creative and sustainable building design.
Pareto Optimization Emphasizing Diversity of Non-Quantifiable Objective

Measures. ArchMOO will utilize various algorithms and methods for Pareto
Optimization, including the most advanced MOGA method: NSGA-II (Deb et al.
2002) to speed up the optimization process, and Interactive Decision Maps (Lotov et
al. 2004) to visualize high dimensional (3 or more objectives) Pareto Optimality.
Diversity of non-dominated solutions is always considered as one of the most
significant aspects in MOGA because crossing over a homogeneous population does
not yield new solutions (Konak et al. 2006). Solution diversity can help avoid
premature convergence or convergence to sub-optimal solutions. A broader
© ASCE
distribution of solutions in the Pareto Optimal Set provides more distinct choices for
designers. In NSGA-II, existing diversity calculation is based on crowding (Deb et al.
2002). In our proposed research, for multiple objectives, building thermal and
daylighting performances will be used as sample quantifiable objectives and the
architectural aesthetics concerning building forms will be used as a sample non-
quantifiable objective. The thermal and daylighting performances will be evaluated
by simulation tools and the aesthetic form needs to be evaluated by architects through
visual inspection.
Form-diversity will help architects focus on the drastically diverse forms to
choose for further design development. In ArchMOO, building form’s shape distance
will be calculated and will work with crowding distances for diversity ranking. Shape
distance calculation will be based on 3D Shape Similarity assessment algorithms
(Shum et al. 1996; Cardone et al. 2003; Bustos et al. 2005; Tangelder & Veltkamp
2008), which compare the 3D shape similarity between closed surfaces. In our
project, since different building design options are generated using parametric
modeling, the causes of the shape changes – parameter and the model graph changes
are known to our system. This information will be used to improve the performance
of 3D shape similarity assessment algorithms. Figure 1 illustrates ideas of Pareto
Optimal solutions with diverse building forms.
Figure 1. ArchMOO ideas illustrated. Left: Pareto Optimal solutions with diverse
building forms (shown as building masses) but similar building forms eliminated
(shown as dark dots); Right: four sample diverse building forms in the Pareto
Optimal solutions displayed in BIM for architects’ inspection and decision-making.
ArchMOO will take advantages of the designers’ intervention during the

optimization process in order to promote creativity through designers’ critical design
thinking, while decreasing the total running time. The optimization framework will
also take advantages of a Cloud-based parametric simulation and optimization
platform (Rahmani Asl et al. 2013, 2014, and 2015), improved optimization methods
based on architectural domain knowledge (Su and Yan, 2014 and 2015), and a new
user interface for parametric architectural design reviews (Altabtabai and Yan 2015).
We have previously built a BIM-based parametric modeling and energy simulation
tool Revit2GBSOpt (Revit to Green Building Studio Optimization) (Rahmani Asl et
al. 2013 and 2014) (Figure 2) and an open source, visual programming and BIM-
based optimization engine – Optimo (Rahmani Asl et al. 2015) (Figure 3).
© ASCE
Revvit2GBSOptt and Optim mo will be incorporateed into ArcchMOO for generating

dessign options parametricaally and evaaluating the solutions inn Pareto Opttimality for
quaantifiable obbjectives - thermal and d daylightinng performaances, while the non-
quaantifiable objjective - form
m-based aessthetics will bbe evaluatedd by architeccts.
Fig
gure 2. Priorr work on BIM
B and Clo
oud-based peerformance optimizationn (Rahmani
Asll et al. 2013 and 2014 4). Left: BIM
M of a testt building. Middle: 2DD graph for
parrametric BIMM. Right: Cloud-based
C thermal annd daylightiing simulatiion runs in
Greeen Buildingg Studio. Th
he project op
ptimizes twoo windows’ design for tthermal and
day
ylighting perrformances.
Fig
gure 3. Optiimo version 0.1.2 graph h for the Scchaffer test ffunction (left), and the
gen
nerated Paretto Optimal set
s (right) (R
Rahmani Asl et al. 2015)..
CO
ONCLUSIONS AND FU
UTURE WO
ORK
Based on the meethodology, our researrch activitiees include ArchMOO

inv
vestigation, prototyping,
p and evaluattion. Becausee of the highh complexityy, currently
a global
g optim
mization for sustainable building deesign cannott be achieveed within a
reasonable timmeframe durring the dessign processs, thus designers’ inteervention is
neeeded to mak ke critical an nd creative decisions inn order to ddirect the ooptimization
tow
wards desired d directions.. The processs of buildinng energy opptimization-bbased form-
find
ding and deesigners’ aesthetic form m-making beecome an appplication aand also an
evaaluation appaaratus for ArrchMOO.
We willl develop a new
n MOGA method bassed on the NSGA-II algoorithm (Deb
et al.
a 2002) by adding non n-quantifiable objectivess that are evaaluated by aarchitects in
the GA processs. ArchMO OO will be an iterative,, multiple L Level of Deetail (LOD)
ocess as follo
pro ows:
1) Architects initializee building design
d accorrding to thee specificatiion using a
paramettric BIM tool,t resultiing in a cconceptual building ddesign with
© ASCE
parametric relationships among the building mass components and external

data such as the site, weather, and sun data. This is a form-making process.
2) Using the current design as a base-design, parametric simulations will run in
the Cloud for thermal and daylighting performance and optimization. In the
optimization, form-diversity will be enforced in NSGA-II for Pareto
Optimality. This is a form-finding process.
3) According to the form diversity-based Pareto Optimal solutions, architects
will choose desired forms based on aesthetics valued by them, as well as
trade-off analyses for those quantifiable objectives. This is a Critical Decision
Point at which undesired solutions are pruned from the decision tree and
computation can be significantly reduced.
4) Design details and manual changes will be made to the forms by architects for
a higher LOD optimization. This again is a form-making process.
5) Repeat #2 until appropriate LOD of the optimal design solutions are reached,
e.g. at the Schematic Design or Design Development levels.
6) Architects will make a post-optimization selection as a final design decision
and further create Construction Documents.
The design process is an iterative form-making and form-finding process. The

Pareto Optimal solutions are linked to the BIM models for visualizing the building
forms (Figure 1).
Based on the research, we will create software prototypes of ArchMOO, which
will be used in experiments, evaluation, teaching integration, and demonstration for
the proposed methods. The design objectives in the prototypes are: maximizing
building thermal and daylighting performances as representative quantifiable design
objectives and form-based aesthetics as a representative non-quantifiable design
objective. ArchMOO will be an open system that allows other objectives to be added
into the system. The proposed design methods will be assessed with quantitative
experiments and qualitative evaluation.
REFERENCES
Altabtabai, J. and Yan, W. (2015). "A User Interface for Parametric Architectural
Design Reviews." Proceedings of the 20th Conference of the Association for
Computer-Aided Architectural Design Research in Asia (CAADRIA).
Kyungpook National University, Daegu, Republic of Korea.
ANSI/ASHRAE (2007). Standard Method of Test for the Evaluation of Building
Energy Analysis Computer Programs.
Bazjanac, V. (2001). "Acquisition of building geometry in the simulation of energy
performance". Building Simulation, 305–311. Rio de Janeiro, Brazil, August
13.
Bradner E, Davis M. (2013). "Design creativity: using Pareto analysis and genetic
algorithms to generate and evaluate design alternatives." Available at:
http://ecologylab.net/workshops/creativity/papers/bradner.pdf
Caldas, L. (2006). "GENE_ARCH: an evolution-based generative design system for
sustainable architecture." Intelligent Computing in Engineering and Architecture.
Springer Berlin Heidelberg. pp.109-118.
© ASCE
Cardone, A., R. K. Gupta, and M. Karnik (2003). "A survey of shape similarity
assessment algorithms for product design and manufacturing applications."
Journal of Computing and Information Science in Engineering.
Coello, C. A. Coello, G. B. Lamont, and D. Van Veldhuisen (2007). Evolutionary
algorithms for solving multi-objective problems, Springer.
Deb, Kalyanmoy, et al. (2002). "A fast and elitist multiobjective genetic algorithm:
NSGA-II. Evolutionary Computation." IEEE Transactions on 6.2 (2002):
182-197.
Gerber, D. J., Lin, S. H. E., Pan, B. P., & Solmaz, A. S. (2012). "Design optioneering:
Multi-disciplinary design optimization through parameterization, domain
integration and automation of a genetic algorithm." Proceedings of the 2012
Symposium on Simulation for Architecture and Urban Design (p. 11). Society
for Computer Simulation International.
Holm, I. (2006). Ideas and Beliefs in Architecture and Industrial design: How
attitudes, orientations, and underlying assumptions shape the built
environment. Oslo School of Architecture and Design. ISBN 82-547-0174-1
Kellert, S. R., J. Heerwagen, and M. Mador (2011). Biophilic design: the theory,
science and practice of bringing buildings to life. John Wiley & Sons.
Konak, A., D. W. Coit, and A. E. Smith (2006). "Multi-objective optimization using genetic
algorithms: A tutorial." Reliability Engineering & System Safety 91.9 (2006): 992-
1007.
LaHood, S., and M. Vanden Brink (2009). "Aesthetics and new product
development." Evidence-based Design for Healthcare Facilities (2009): 19.
Lotov, A. V., Bushenkov, V. A., & Kamenev, G. K. (2004). Interactive decision
maps: Approximation and visualization of Pareto frontier (Vol. 89). Springer.
Pollio, V. (1914). Vitruvius: The Ten Books on Architecture (Vol. 645). Courier
Dover Publications.
Radford, A. D., and J. S. Gero (1987). Design by optimization in architecture,
building, and construction. John Wiley & Sons, Inc.
Rahmani Asl, M., Zarrinmehr, S., and Yan, W. (2013). "Towards BIM-based
Parametric Building Energy Performance Optimization." Proceedings of The
Association for Computer Aided Design in Architecture (ACADIA), October
24-27, 2013. Cambridge, Ontario, Canada.
Rahmani Asl, M., Bergin, M., Menter, A., and Yan, W. (2014). "BIM-based
Parametric Building Energy Performance Multi-Objective Optimization."
Proceedings of the Conference of Education and Research in Computer Aided
Architectural Design in Europe (eCAADe), Northumbria University,
Newcastle Upon Tyne, UK.
Rahmani Asl, M., Stoupine, A., Zarrinmehr, S., and Yan, W. (2015). "Optimo: A
BIM-based Multi-Objective Optimization Tool Utilizing Visual Programing
for High Performance Building Design." Proceedings of the Conference of
Education and Research in Computer Aided Architectural Design in Europe
(eCAADe), Vienna, Austria.
Shum, H., M. Hebert, and K. Ikeuchi, (1996). "On 3D shape similarity." Proceedings
of Computer Vision and Pattern Recognition, CVPR'96, IEEE Computer
Society Conference on. IEEE.
© ASCE
Su, Z., Yan, W. (2014). "Improving Genetic Algorithm for Design Optimization
Using Architectural Domain Knowledge." Proceedings of the Annual
Conference of the Association for Computer Aided Design in Architecture
(ACADIA), October 23-25, 2014. Los Angeles, California.
Su, Z. and Yan, W. (2015). "A Fast Genetic Algorithm for Solving Architectural
Design Optimization Problems." Artificial Intelligence for Engineering
Design, Analysis and Manufacturing (AIEDAM), Cambridge University Press.
Accepted.
Tangelder, J. W., & Veltkamp, R. C. (2008). "A survey of content based 3D shape
retrieval methods." Multimedia tools and applications, 39(3), 441-471.
Torcellini, P., Pless, S., Deru, M., Griffith, B., Long, N., & Judkoff, R. (2006).
Lessons learned from case studies of six high-performance buildings, National
Renewable Energy Laboratory.
Welle, B. M. Fischer, J. Haymaker, and V. Bazjanac (2012). "CAD-Centric
Attribution Methodology for Multidisciplinary Optimization (CAMMO):
Enabling Designers to Efficiently Formulate and Evaluate Large Design
Spaces." CIFE Technical Report #195, Center for Integrated Facility
Engineering, Stanford, CA.
© ASCE
Inequality as a Barrier to Green Building Policy Adoptions in Cities
Aaron Deslatte1; Kathryn Wassel2; and Richard C. Feiock2

1
Northern Illinois University, Department of Public Administration, IASBO Building
(2nd Floor), DeKalb, IL 60115. E-mail: adeslatte@niu.edu
2
Florida State University, Askew School of Public Administration and Policy, 627
Bellamy Building, Florida State University, Tallahassee, FL 32306. E-mail:
kathryn.wassel@gmail.com; rfeiock@fsu.edu
Abstract
How might social inequality influence local government decisions to adopt

green building incentives or mandates? Extant research has considered the influence
that political homophily, environmental activism, and pressure groups play in the
spread of sustainability policies. But to what extent do concerns over environmental
justice motivate local government adoption of green building, development, and
remodeling programs? This study begins to disentangle these questions by asking: do
cities within regions with higher levels of social inequality adopt green building
policies less frequently than those in areas of lower inequality? Using national survey
data, we find measures of wealth and “reformed” government positively influence the
use of green-building tools, while regional income inequality dampens these effects.
INTRODUCTION
Urban planners face a distributional dilemma in the modern metropolis. Green

building policy tools tend to be adopted by wealthier, more racially homogenous
communities in the United States. Ironically, these are the communities which rarely
need innovation to attract commercial activity, redevelop blighted and
environmentally degraded areas, or meet the demands of vulnerable populations.
Greater income inequality and racial/ethnic segregation are often associated with the
development pattern known as urban sprawl, as well as the diminishment of social
capital. Extant social science research has considered the influence that political
homophily, environmental activism, and development interests play in the spread of
sustainability policies generally (Feiock, Tavares and Lubell 2008; Gerber, Henry and
and Lubell 2010; Krause 2010). While this research largely finds measures of
affluence and education influence policy diffusion, they generally do not directly
account for income stratification within metropolitan regions. This study begins to fill
this gap directly by asking: do cities nested within regions with higher levels of social
inequity adopt green building policies less frequently than those with lower levels of
social equity?
Using a large sample of U.S. cities, we use survey and archival data to
investigate a range of demographic and socio-environmental drivers of green building
© ASCE
policy adoptions by local governments. The next section explains extant theory and
research on green policy tools, measurements of social equity, and presents
hypotheses for their association. The third section details our data and analytic
approach. The fourth section presents results, and the concluding section speaks
briefly about extensions for this research.
GREEN BUILDING POLICY TOOLS AND SOCIAL EQUITY
Policy tools research has a long tradition in economics based on market

failures and government intervention. Environmental policy tools research has
evolved from a focus on top-down coercive methods for curbing pollution to more
market-based mechanisms and voluntary approaches in recent decades (Fiorino 2006;
Keohane, Revesz and Stavins 1998). The study of land-use policy tools has also
transitioned to focusing on urban containment strategies which facilitate growth
rather than attempting to control or block it (Anderson 1998). Urban planners,
political scientists, and public administrationists alike have shifted their focus from
exclusionary zoning, growth controls and state-directed comprehensive planning to
“smart growth” strategies aimed at revitalizing blighted urban areas while attempting
to preserve natural amenities without exacerbating social inequities (Chapin 2012;
Feiock, Tavares and Lubell 2008; Hawkins 2011). This transition has occurred in
tandem with a heightened recognition that future human settlement patterns in
metropolitan areas will play a major role in global efforts to reduce greenhouse gas
emissions (Howell-Moroney 2007; IPCC 2014).
Local governments have adopted a range of policy tools to encourage green
building practices, including elements of life cycle assessment (LCA) for
scientifically evaluating the environental impacts of buildings; incentives for energy
efficiency through high performance design and solar power; increasing density in
land uses; water conservation; and utilizing codes for achieving green building
certifications such as GreenStar in Austrailia, Green Home Evaluation Manual
(GHEM) in China, or Leadership in Energy & Environmental Design (LEED) in the
United States (Ding 2008). Specific green building policy tools available to local
governments include: energy audits of homes and businesses; weatherization
programs; incentives for HVAC upgrades, energy efficient appliances, and solar
equipment; requiring new government building construction or retrofitting to be
LEED or Energy Star certified; permiting higher density developmnt near public
transit nodes; providing density incentives; reducing fees for environmentally
friendly development; and zoning which encourages mixed-use development.
Green Building Policy Tools. Green building policies are often pursued by
communities with the administrative and material means to do so. This presents a
green policy paradox of sorts. Communities which may be in greatest need of green
investment – lower-income, blighted and higher minority communities with
deteriorating infrastructure and services – should generally be less capable of
inducing development interest to make the investments necessary to spur green policy
tool utilization. In other words, cities with fewer resources and less development
demand for innovative building designs and energy efficiencies are generally less
© ASCE
likely to take governmental or community-wide green building actions. At the same

time, income inequality has steadily increased since 1981 as measured by the U.S.
Census Bureau via its GINI index capturing spatial disbursion of household income.
If policy networks and collaboration play a role in diffusion, the diminishment of
potential peers or partners in green-building innovation poses a potential barrier to it.
Extant research does not provide an answer as regional inequality has not been
explored emprically as a driver or barrier to policy adoption.
There may be several reasons for a relationship. Increasing neighborhood
inequality fuels population sorting in which the social capital, resources and positive
economic spillovers of higher-income households become segregated within smaller
pockets of metropolitan regions, abandoning other quadrants to poorer, often racial-
minority concentrated populations (Hardman and Ioannides 2004). This urban
phenomenon of income and racial stratification has come to define the American
urban landscape (Downs 2003). As a result, researchers have devoted decades to
studying the dimensions of citizens “voting with their feet” and its effect on human
behavior, capacity for innovation, and willingness of government and business actors
to coordinate (Anthony 2004; Carr and Tavares 2014; Downs 1999; Epple and
Zelenitz 1981; Fischel 2006; Howell-Moroney 2008).
While we expect that income and education have the most powerful
stratifiying influence on green policies, we also expect this effect is moderated by the
level of income inequality -- as measured by the Census GINI coefficients – present
at the metroplitian level. Our analysis utilizes measures of median home value and
median household income as indicators of wealth at the city level, and we follow the
lead of Berry et al., (2012) to develop conditional propositions and test interaction
hypotheses.
We expect that due to population sorting along income lines, metropolitian
areas with higher income and poverty stratification in different cities will
cumulatively adopt fewer numbers of green builidng tools -- that is, metro-level
income inequality will “sort out” the demanders of green building policies.
Metropolitan areas with lower amounts of income inequality could conceivably
witness less political and social homophily and a larger proportion of municipalities
adopting green policy tools. However, we expect that a community’s relative degree
of wealth and income will still influence the propensity of its government to pursue
green building tools even in less income-stratified regions. We use median home
values as a measure of wealth since local governments derive their revenues primarily
from property taxes rather than income or services taxes. Thus:
Proposition 1: The marginal effect of home values will be positively
associated with the adoption of greater numbers of green building policy tools when
metro inequality is at its lowest level.
As inequality rises, we expect social capital and positive spillovers from
mixed-income neighborhoods to diminish. Thus, it follows that the influence of
wealth on policy tool adoption will be reduced in higher-income communities when
the region’s income is more unequally dispersed between cities.
Proposition 2: The marginal effect of home values will be negatively
associated with the adoption of greater numbers of green building policy tools when
metro inequality is at its highest level.
© ASCE
These two wealth indicator propositons lead to the first interaction hypothesis:
Hypothesis 1: The marginal effect of wealth indicators on green building

policy tool adoption is positive at lower levels levels of metropolitan
inequality; this effect decreases as metro inequality increases and eventually
becomes negative.
Since more informaton on the long-term benefits would also alleviate such
cost concerns, communities with higher levels of education among its population
could also be more likely to adopt them.
Proposition 3: The marginal effect of community education levels will be
positively associated with the adoption of greater numbers of green building policy
tools when income inequality is at its lowest level.
Proposition 4: The marginal effect of community education levels will be
positively associated with the adoption of greater numbers of green building policy
tools when income inequality is at its highest level.
Hypothesis 2: The marginal effect of community education levels on green

building policy tool adoption is positive at all levels of metropolitan
inequality; this effect grows larger as metro inequality increases.
Public administration and political science scholars also have a long tradition
of considering how the strucutre of local government decision-making influences
policy choices (Carr 2015; Clingermayer and Feiock 2001). A central argument for
why the Council-Manager form of government may be more active in pushing to
adopt green-building policies is that they represent “win-win” actions which seek to
grow and green local economies, advancing the career objectives of city managers
which seek to establish track records of innovation as well as environmental
sustainability and growth. We therefore expect that local governments in which
managers hold more functional control over executive authority of the government
will be more actively pursuing such green-building policy tools.
Hypothesis 3: Managerial form of government will be positively associated

with green building policy tool adoption.
Social Inclusion Policy Tools. Although equity has been considered one of the
“three E’s” of sustainability, the ramifications of urban sustainability for the urban
disadvantaged has received little attention from the broader research community
(Berry 2001; Curley 2010). While sustainability proponents cite the need for equity to
be considered along side other efficiency or environmental concerns, it is often
consiered only in certain contexts or subject to tradeoffs. Growing local governments
have been cogniscent of the need to provide workforce housing, advocates have
pushed for socal inclusionary land uses and housing support for the disabled, seniors,
the homeless, while urban reinvestment strategies often tout inclusionary efforts.
A predominant sociological approach holds that material resource dispersal
may be the most noticeable consequence of inequality, and this lack of resources
© ASCE
leads to systematic under-investment in the services and greater disparity across a

wide range of social outcomes including health and civic and political participation.
Country-level research on inequality has failed to reconcile the expectaton that
inequality hastens the demand for redistributive policies with this inconsistent
evidence of such an effect. This may be in part because policymakers are more likely
to be responsive to the portion of the electorate that turns out to vote and can marshal
political resources, while the standard median voter theorem assumes 100%
participation (Werfhorst, et al. 2012). In reality, participation is biased and predictors
such as age and occupation, income, and education have been shown to influence
who votes, and is therefore more likely to have their preferences aggregated into
policy choices.
Another element of inequality is related to racial stratification within
metropolitan regions. Housing segregaton remains pervasive around the United
States, connected to a complex interaction of racial stereotypes and preferences for
homophily, income and labor skill and education inequality (Ihlanfeldt and Scafidi
2004). The federal Housing and Urban Development’s most recent Housing
Discrimination Study (2000) determined that discrimnation still existed in both rental
and sales markets in large U.S. metropolitan areas, although it had decreased in both
categories for African-Americans since 1989. While in decline, we expect that the
cumulative effect of decades of racial housing preferences and exclusionary policies
still influences the motivation of localities to invest in socially inclusive housing
policies. We also expect this negative relationship will be stronger when income
inequality is greater. We use measures of the percentage of white citizens, the median
home value in cities, the rate of homeownership, median home value and median
family income to examine the pollowing propositions:
Proposition 5: The marginal effect of measures of housing segregation will be
negatively associated with the adoption of social inclusion policy tools when income
inequality is at its lowest level.
Proposition 6: The marginal effect of measures of housing segregation will be
negatively associated with the adoption of social inclusion policy tools when income
inequality is at its highest level.
These form our interaction hypothesis that metropolitan inequality
exacerbates the reluctance of municipalities to adopt what are in essence
redistributional policies aimed at assisting the homeless, elderly, and disabled.
Hypothesis 4: The marginal effects of housing segregation measures such as

race, homeownership rates and income on social inclusion policy tool
adoption will be negative at all levels of metro inequality; this negative effect
will strengthen as inequality increases.
DATA AND METHODOLOGY
To analyze policy tool adoption, we utilize a national survey of local

governments administered by the International City/County Management Association
(ICMA) in 2010. The survey was sent to all municipalities with populations larger
than 2,500, asking a variety of questions related to their form of government,
© ASCE
demographics and sustainability practices. The survey response rate was 25.4%, with
2,176 cities and counties responding. We restrict the responses to cities only, yielding
1,867 observations about adoption of 56 questions pertaining to adoption of
greenhouse-gas emissions, smart growth, energy efficiency, and social
inclusion/equity policy tools. Our measures of socio-economic, demographic and
governmental form are derived from the 2010 U.S. Census and the survey responses.
Standard practice in sustainability research to date has been to create single
additive indexes of all sustainability tools or to attempt to differentiate between types
of tools with distinctive political “marketplaces” of demanders and suppliers within
communities. In their 2013 study using the same ICMA survey, Opp and Saunders
(2012) utilized 84 “indicators” of environmental policy, sustainable development, and
social equity tools to create a single measure (the Opp-Suanders Sustainability
Practices Index, or OSSPI) which attempted to capture the three “E’s” within a single
weighted index. This index is composed of three underlying subindices which proport
to capture the willingness of local governments to support environmental, economic
development, and equity causes. Thus, the higher a city scored on an range of 0 to
100, the more “sustainable” it would be considered.
We adopt a different approach to the question of whether the “Three E’s” are
arbitrary classifications of policy choices. First, we use factor analysis to disentangle
five distinct types of green policy tools which tend to be lumped together:
Greenhouse gas reduction targets; energy efficiency; green building; land use; and
social inclusion. We then fit bootstrapped zero-inflated negative binomial models for
the green building and social inclusion indexes to test the conditional hypothesis that
concentrations of wealth, education, fiscal resources and form of government within
cities interact with regional inequality to influence their tool adoptions.
This analysis first compared the performance of these concepts to factor
scores which attempt to reduce the number of variables and capture the underlying
latent structures of the policy tool space. We utilized exploratory factor analysis
(EFA) to determine whether a multivariate strategy for variable reduction can account
for more of the covariance in policy tool adoption and more accurately identify the
underlying latent concepts of sustainable development and equity which are of
theoretical interest. The goal of factor analysis to identify latent variables which may
cause manifest variables to co-vary (Floyd and Widaman 1995). Because the vast
majority of the policy tool responses are not normally distributed, we utilized a
principal factor extraction method and orthogonal Varimax rotation, which identified
five factors through a scree test which had eigenvalues > 1 and minimal, if any,
crossloadings (Costello and Osborne 2005). The five retained factors roughly
correspond to the five functional sustainability policy categories included in the
survey: greenhouse-gas reduction goals; energy efficiency measures; land-use; green-
building; and social inclusion tools. This suggests there are likely underlying distinct
causal factors for their adoption. Rather than merging all tools, we examine the count
of polices adopted for the dimensions of interest: green-building and social-inclusion
tools.
Outcome measures. The green building index was zero-inflated, containing 12

possible tools local governments could have adopted: permitting higher density
© ASCE
development near public transit nodes; permitting higher density development in

locations where infrastructure such as utilities are already in place; incentives other
than density for new commercial development that are LEED certified or an
equivalent; incentives other than density for new single-family residential that is
LEED certified or an equivalent; apply LEED Neighborhood Design standards;
provide density incentives for “sustainable” development; provide tax incentives for
“sustainable” development; reduce fees for environmentally friendly development;
fast track plan reviews and/or inspections for environmentally friendly development;
residential zoning codes to permit solar installations, wind power, or other renewable
energy production; residential zoning codes to permit higher densities through
ancillary dwelling units or apartments (basements, garage units, in-house suites);
zoning codes to encourage mixed-use development.
The social equity index was also zero-inflated and contained 16 potential
social inclusion policy tools: providing financial support/incentives for affordable
housing; supportive housing to people with disabilities; housing options for the
elderly; housing to the homeless; access to information technology for persons
without an internet connection; funding for pre-school education; after-school
programs for children; report on community quality of life indicators, such as
education, cultural, diversity, and social well-being; actions through either restrictions
or incentives to use of locally grown produce; restrictions on the purchase of bottled
water by the local government; use of public lands for community gardens; support
for local farmer's markets; community education programs dealing with the
environment or energy conservation; locating recycling containers close to refuse
containers in public spaces such as streets and parks; and adopting a green purchasing
policy for the local government.
Inequality measure. Inequality is measured at the Metropolitan Statistical Area

(MSA) level for each municipal survey respondant within one. With Census GINI
measures for 2007 and 2012, we took the mean of these two points in time in order to
capture a more accurate measure corresponding to the 2010 time period of the survey.
. Linear OLS models demonstrated serious violations of Gauss-Markov assumptions
of linearity and homoskedasticity, therefore we used negative-binomial models.
Because roughly two-thirds of the responses are from cities within micropolitan
statistical areas or rural enough to be considered outside any such region, we omit
these observations from our interaction models, which drops the number of
observations for those two models to 616 cities. Although numerous statistical
analyses have been done over the decades with similar truncation of the data (whch
effectively limits the same to cities in metros of 1 million people or more) we opted
to also include two non-interaction models capturing only the direct effects of the
explanatory variables but retaining the full sample of city responses. We used
bootstrapped standard errors to address potential bias. We also opted to not utilize a
multilevel model estimation even though the regional data are clearly hierarchical in
nature. This was because a large number of metro areas represented within the sample
fell below having enough Level 1 responses (<5) to properly fit a multilevel model.
This is not an ideal solution, but was necessary due to the quality of data. Table 1
describes the indexes and explanatory variables in the models and data sources.
© ASCE
We fit four models – two measuring the additive influence, and two
interaction models which account for regional inequality -- to gauge the drivers of
green building and social equity-related policy tool adoptions nationally. Tables 2 and
3 in the following section display the results.
Table 1 Descriptive Statistics

Outcome Variables Source Obs Mean SD Min Max
Green Building Index ICMA Survey 1,769 1.47 1.9 0 11
Equity Index ICMA Survey 1,769 3.43 3.05 0 14
Explanatory Variables
Median Age 2010 U.S. Census 1,844 38.25 6.23 21 70.9
% White 2010 U.S. Census 1,844 80.7 17.38 3.3 99.3
% W/Bachelor's or higher 2010 U.S. Census 1,633 27.97 16.11 3.3 86.1
Median home value 2010 U.S. Census 1,633 $221,016 $180,816 $28,200 $1,000,001
Homeownership 2010 U.S. Census 1,844 66.18 13.37 20.3 97.5
Median Income 210 U.S. Census 1,633 $66,549 $28,607 $23,690 $250,001
GINI Average U.S. Census 730 0.464 0.018 0.426 0.505
Per Capita Property tax U.S. Census 1,808 $428 $1,233 $0 $47,981
Manager ICMA Survey 1,867 0.574 0.495 0 1
Commission ICMA Survey 1,867 0.013 0.117 0 1
age*GINI U.S. Census 722 18 2.94 9.26 33.25
White*GINI U.S. Census 722 36.44 8.34 1.44 48.98
Education*GINI U.S. Census 627 15.84 8.39 2.04 41.62
Home Value*GINI U.S. Census 627 139974 101734 21929 504450
Income*GINI U.S. Census 627 37156 15430 12734 112627
Homeowner*GINI U.S. Census 722 32.45 6.73 10.86 47.87
Population (logged) 2010 U.S. Census 1,844 9.49 1.16 4.3 15.92
% Pop Change 2000, 2010 Censu 1,833 16.26 100.45 -71.5 4081
Pop per sq mile (logged) U.S. Census 1,844 7.29 0.937 0.742 10.2
RESULTS
We find evidence that social stratification within metropolitan areas broadly

contributes to the variation in municipal adoption of both green-building tools, but
less evidence that it impacts decisions about social inclusion policies. While this is a
useful first step in analysis, causal inference remains a problematic issue with cross-
sectional data and the exact causal mechanisms remain under-identified in our
models. We now discuss the specific evidence relevant to our theoretical
expectations.
Hypothesis 1: The marginal effect of wealth indicators on green building policy tool
adoption is positive at lower levels levels of metropolitan inequality; this effect
decreases as metro inequality increases and eventually becomes negative.
In the green-building model, we find evidence that median home values have
a positive direct influence on green building policy tool adoption but that this effect
lessens and actually turns negative when regional inequality is higher. The results
generally conform to our expectations that more affluent communities are more likely
© ASCE
to innovate by offering incentives or regulatory breaks for density-increases, LEED

certified projects, or other sustainability-related construction projects. This supports
the paradox that the communities most in need of such innovations are less likely to
engage in them. This stratification effect is not linear, however. The statistically
significant, negative coefficients for the GINI interaction terms with both race and
home values suggests the marginal effects of these influences declines within
metropolitan regions where social inequality is higher. A possible explanation for this
finding is that communities in which wealth is concentrated in fewer hands – and the
population is more stratified around racial or ethnic lines – leads to still greater
diminishment in what sociologists think of as the social capital necessary to pursue
broader environmental programs and both richer and poorer communities are less
likely to engage in sustainability efforts. Even green building programs, which have
demonstrated themselves to be profitable for the private sector and often pay for
themselves in terms of government investment, suffer. Innovative cities within metros
with higher inequality may also find it more difficult to locate like-minded partners to
pursue collaborations, spark competition for green jobs, or share experiences via
information networks.
Hypothesis 2: The marginal effect of community education levels on green building

policy tool adoption is positive at all levels of metropolitan inequality; this effect
grows larger as metro inequality increases.
We find evidence that education levels influence green policy tool adoption
directly, but the relationship is more complicated and once moderated by inequality
regionally. We expected higher levels of education to facilitate the diffusion of green-
policy tools by reducing the information costs for voters and builders, and thus
leading to increased acceptance or even demand for such policies. Our non-
interactive model appears to conform to this expectation, with a positive statistically
significant effect. This relationship becomes more complex once the interaction effect
is considered. We find the direct effect has now become negative, while the
interaction of education levels and income inequality shifts this negative slope back
into a positive direction as inequality increases. Thus, the suggestion is that education
levels have a greater positive influence on green-building adoption in metros where
income inequality is greater.
Hypothesis 3: Managerial form of government will be positively associated with

green building policy tool adoption.
Conforming to our expectation, managerial form of government is positively
associated with green building policy tool adoption. Council-manager forms of
government have been touted for more than century as a more efficient, effective and
politically insulated manner in which to administer city departments and impliment
public policies. While the findings surrounding form of government have been mixed
related to sustainability, this is likely due to the tendency of researchers to lump all
sustainability tools into an equally weighted index in whih the tools managers may
have more professional and career incentives to pursue – those which will look good
in marketing themselves for their next job, or improve the fiscal condition of the city
– are lost in the noise of policies in which strong political advocacy or constituent
© ASCE
demand are more likely to lead to their adoption. We now turn to the social inclusion
model.
Table 2 Gre e n Building Tool Count Mode ls

Explanatory Variables Model 1 (Bootstrap S.E.) Model 2 IRR
Median Age -.225 (.006) ** -.049 (.206)

% White -.004 (.002) ** .215 (.07) ** 1.24
% w/Bachelor's or higher .015 (.004) ** -.18 (.09) * 0.836
Median home value .000002 (.0000003) ** .00002 (.000008) * 1.00002
Homeownership .008 (.004) ** .004 (.005)
Median Income -.000013 (.000004) ** -.000009 (.000005)
GINI Average 26.84 (15.7)
Per Capita Property tax .00006 (.00007) -.00001 (.00009)
Manager .275 (.074) ** .309 (.1) ** 1.36
Commission -.198 (25.22) -33.87 (.0000001)
age*GINI .07 (.45)
White*GINI -.48 (.15) ** 0.622
Education*GINI .41 (.2) * 1.51
Home Value*GINI -.00004 (.00002) * 0.9999
Constant 1.08 (.216) ** -10.95 (7.22)
Inflation Model (Logit)

Population (logged) -.679 (.129) ** -.84 (.2) ** -.84
% Pop Change -.0295 (.009) ** -.018 (.01)
Pop per sq mile (logged) .065 (.134) -.031 (.2)
Constant 4.93 (1.15) ** 7.16 (2.002) ** 7.16
Wald chi2 120.37 61.26
Prob > chi2 0.000 0.000
N 1,604 616
**p < .01, *p<.05
Hypothesis 4: The marginal effects of housing segregation measures such as race,

homeownership rates and income on social inclusion policy tool adoption will be
negative at all levels of metro inequality; this negative effect will strengthen as
inequality increases.
While the non-interactive model estimation for social inclusion finds evidence
that housing segregation measures such as the percentage of whites, household
income and the rate of homeownership are negatively associated with adoption of
social equity tools, this effect washes out in the interaction model. In fact, once
metro-wide inequality is included in the model, we also lose positive statistically
significant effects for education and home values. The interaction model finds
support for previous research which has posited that city size, per capita reliance on
property taxes for revenue, and managerial form of government may positively
influence the likelihood of adopting a higher number of inclusionary housing, urban
revitalization, or other equity tools.
© ASCE
Table 3 Social Inclusion Tool Count Mode ls

Explanatory Variables Model 3 (Bootstrap S.E.) Model 4 IRR
Median Age -.009 (.004) * .158 (.145)

% White -.002 (.001) * .038 (.049)
% w/Bachelor's or higher .011 (.002) ** -.051 (.092)
Median home value .000002 (.0000002) ** .0000006 (.000008)
Homeownership .008 (.004) ** .006 (.091)
Median Income -.00001 (.000002) ** .00003 (.00008)
GINI Average 23.9 (11.71) * 2.40E+10
Per Capita Property tax .0002 (.00006) ** .0002 (.00007) ** 1.0002
Manager .148 (.046) ** .166 (.074) * 1.18
Commission -.057 (.258) -.038 (.375)
age*GINI -.377 (.317)
White*GINI -.092 (.106)
Education*GINI .135 (.199)
Home Value*GINI .000002 (.00002)
Income*GINI -.00008 (.0002)
Homeowner*GINI -0.027 (.195)
Constant 2.24 (.128) ** -8.44 (5.43)
Inflation Model (Logit)

Population (logged) -1.43 (.231) ** -1.37 (.393) ** -1.37
% Pop Change .0005 (.005) .0002 (.001)
Pop per sq mile (logged) .246 (.188) .067 (.236)
Constant 8.73 (1.97) ** 9.876 (3.65) ** 9.87
Wald chi2 (9) 314.11 140.24
Prob > chi2 0.000 0.000
N 1604 616
**p < .01, *p<.05
The largest positive driver of adoption in the model is the direct measurement
of income inequality, which suggests metro areas which experience higher degrees of
housing stratification based on income are more likely to generally adopt higher
numbers of inclusionary policies. This suggestion makes sense intuitively, but
without modeling inequality as a hierarchical fixed or random effect, we hesitate to
draw such a conclusion based of this analysis. In other words, we find evidence only
that larger cities with greater property-tax bases – cities which are also more likely to
have larger populations in need of these services -- are more likely to adopt larger
numbers of these policies.
CONCLUSIONS
We are only beginning to understand the role that social cohesion across city-
limits plays in the adoption of policies which benefit human building ecosystems.
Environmental protection, economic development and social equity are distinct terms
co-branded as “sustainability” in the academic literature and often treated as if they
are similar investments. But disentangling the “why?” of these separate dimensions of
governmental action requires unraveling the distinct political economies and drivers
© ASCE
of differently weighted policy tool adoption, as well as the lnkages between cities
within metropolitian regions.
This paper offers one step in that goal by demonstrating first that the basket of
“sustainability” tools frequently lumped together with equal weight in extant
scholarship manifest from different objectives and should be modeled with great care
– and likely individually from one another. Green building tools are likely valued
differently across local governments, sectors of the economy and citizen subgroups
from socially inclusive policy goals even when they also promote sustainability.
This research has much room for advancement. Besides the causation
problem, there are likely omitted variables such as interest-group pressure which our
models for now leave out. Better theory and modeling of why region-wide social
disparities might influence intra-governmental decisions is needed. As a second
phase, we intend to conduct a comparative case analysis of these drivers as well as
gathering richer health-related outcome and building condition data across cities in
the Chicago and Orlando metropolitan regions.
BIBLIOGRAPHY
Berry, William D, Matt Golder, and Daniel Milton. 2012. "Improving Tests of
Theories Positing Interaction." Journal of Politics 1-19.
Bickers, Kenneth N, Lapo Salucci, and Robert M Stein. 2006. "Assessing the Micro-
Foundations of the Tiebout Model." Urban Affairs Review 42-57.
Boyle, Robin, and Rayman Mohamed. 2007. "State Growth Management, Smart
Growth and Urban Containment: A Review of the US and a Study of the
Heartland." Journal of Environmental Planning and Management 677-697.
Bromley, Daniel W. 1991. Environment and the Economy: Property Rights and
Public Policy. Cambridge, MA: Basil Blackwell.
Brown, Trevor L., and Matthew Potoski. 2003. "Transaction Costs and Insitutional
Explanations for Government Service Production Decisions." Journal of
Public Administration Research and Theory 441-468.
Carr, Jared B, and Antonio Tavares. 2014. "City Size and Political Particiaption in
Local Government: Reassessing the Contingent Effects of Residential
Location Decisions within Urban Regions ." Urban Affairs Review 269-302.
Carr, Jared. 2015. "What Have we learned about the Performance of Council-
Manager Governent: A Review and Synthesis of the Research Literature."
Public Administration Review Forthcoming.
Chapin, Timothy. 2012. "From Growth Controls, to Comprehensive Planning, to
Smart Growth: Planning’s Emerging Fourth Wave." Journal of the American
Planning Association 5-15.
Clingermayer, James, and Richard Feiock. 2001. Institutional Constraints and Policy
Choice: An Exploration of Local Governance. Albany: State University of
New York Press.
Curley, Alexandria M. 2010. "Relocating the Poor: Social Capital and Neighborhood
Resources." Journal of Urban Affairs 79-103.
deHaven-Smith, Lance. 2000. "Facing Up to the Political Realities of Growth
Management." Florida Planning 14-17.
© ASCE
Downs, Anthony. 1999. "Some realities about sprawl and urban decline." Housing
Policy Debate 955-74.
Feiock, Richard C., Antonio F. Tavares, and Mark Lubell. 2008. "Policy Instrument
Choices for Growth Management and Land Use Regulation." The Policy
Studies Journal 461-480.
Finseraas, Henning. 2009. "Income Inequality and Demand for Redistribution: A
Multilevel Analysis of European Public Opinion." Scandinavian Political
Studies 94-119.
Fiorino, Daniel. 2006. The New Environmental Regulation. Cambridge: MIT Press.
Fischel, William A. 2006. The Tiebout Model at Fifty. Cambridge: The Lincoln
Institute of Land Policy.
Floyd, F.J., and K.F. Widaman. 1995. "Factor analysis in the development and
refinement of clinical assessment instruments." Psychological Asessment 286-
299.
Gerber, Elisabeth R., Adam Douglas Henry, and Mark and Lubell. 2010. "The
Political Logic of Local Collaboration in Regional Planning in California.
Paper 35." Duke Networks Conference. Accessed 1 30, 2014.
http://opensiuc.lib.siu.edu/pnconfs_2010/35.
Hardman, Anna, and Yannis M. Ioannides. 2004. "Neighbors’ Income Distribution:
Economic Segregation and Mixing in US Urban Neighborhoods." Journal of
Housing Economics 368–82.
Hawkins, Christopher V. 2011. "Smart Growth Policy Choice: A Resource
Dependency and Local Governance Explanation." Policy Studies Journal 679-
707.
Heyward, M. 2007. "Equity and international climate change negotiations: A matter
of perspective." Climate Policy 518–534.
Howell-Moroney, Michael. 2007. "Studying the Effects of the Intensity of the U.S.
State Growth Management Approaches on Land Development Outcomes."
Urban Studies 2163-2178.
Howell-Moroney, Michael. 2008. "The Tiebout Hypothesis 50 Years Later: Lessons
and Lingering Challenges for Metropolitan Governance in the 21st Century."
Public Administration Review 97-109.
Ihlanfeldt, Keith R, and Benjamin Scafidi. 2004. "Whites' neighbourhood racial
preferences and neighbourhood racial composition in the United States:
evidence from the multi‐city study of urban inequality." Housing Studies 325-
359.
IPCC. 2014. Intergovernmental Panel on Climate Change Working Group III
contribution to Fifth Assessment Report. United Nations, New York: IPCC.
Jun, Hee-Jung. 2013. "Determinants of Neighborhood Change: A Multilevel
Analysis." Urban Affairs Review 319-352.
Kang, In-Sung, and Richard C Feiock. 2006. "Implementation of Growth
Management Policy in Florida Cities: Zoning Approval and Regulatory Policy
Enforcement." International Review of Public Administration 85-98.
Keohane, N.O., R.L. Revesz, and R.N. Stavins. 1998. "The Choice of Regulatory
Instruments in Environmental Policy." Harvard Environmental Law Review
313-68.
© ASCE
Klinsky, Sonja, and Harald Winkler. 2014. "Equity, sustainable development and
climate policy." Climate Policy 1-7.
Krause, Rachel M. 2011. "Symbolic or Substantive policy? Measuring the extent of
local commitment to climate protection." Environment and Planning C:
Government and Policy 46-62.
© ASCE

Sustainable Human-Building Ecosystems: Yimin Zhu, Khee Poh Lam, Yong Tao

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Sustainable Human-Building Ecosystems: Yimin Zhu, Khee Poh Lam, Yong Tao

Uploaded by

Copyright:

Available Formats

Sustainable

October 5-6, 2015

Research Coordination Network on Sustainable Human–Building

The Architectural Engineering Institute

Published by the American Society of Civil Engineers

Errata: Errata, if any, can be found at http://dx.doi.org/10.1061/9780784479681

Copyright © 2015 by the American Society of Civil Engineers.

Welcome to the First International Symposium on Sustainable Human–Building

On behalf of the organizing committee, we would like to thank the School of

Human Ecology and Building Science: A Necessary Synthesis .............................. 1

Occupant Behavior Modeling and Analysis

Occupant Behaviors and Energy Use: Creating High-Performance People for

Participatory Energy Management in Building Networks ................................... 27

One Size Does Not Fit All: Eco-Feedback Programs Require

Development of Non-Intrusive Occupant Load Monitoring (NIOLM) in

Default Conditions: A Reason for Design to Integrate Human Factors .............. 54

Assessing Energy Strategies in Active Buildings Considering

Estimating Occupancy in an Office Setting............................................................ 72

Event-Based Parallel Simulation with a Sensing System for Occupant

Effects of Variant Occupancy Transitions on the Energy Implications of

Thermal Comfort Prediction and Analysis

A Study of Time-Dependent Variations in Personal Thermal Comfort via a

Occupant Individual Thermal Comfort Data Analysis in an Office .................. 108

Facial Skin Temperature as a Proactive Variable in a Building Thermal

Advancing Occupant-Centered Performance Simulation Metrics Linking

Direct Measurement of Occupants’ Skin Temperature and Human

Innovative Planning, Design, and Policies for Building Energy Efficiency

Incorporation of Future Building Operating Conditions into the Modeling

Measuring the Effectiveness of an Immersive Virtual Environment for the

Green Building Design as If People Mattered ...................................................... 176

Integration of QFD and Utility Theory to Improve End-User Satisfaction

The Power of Data Visualization: A Prototype Energy Performance Map

Using Relationship Mapping to Understand Sustainable Housing

The Weatherization Assistance Program: Social Policy or Energy Policy

Towards Multi-Objective Optimization for Sustainable Buildings with

Inequality as a Barrier to Green Building Policy Adoptions in Cities............... 231

Human Ecology and Building Science: A Necessary Synthesis

Department of Building Physics and Building Ecology,

Buildings represent a major class of interventions by people in the

Ecological potency refers to the human repertoire of means to deal (cope,

sound pressure level) to which subjective judgments (such as the degree of

sustainability considerations play an important role also in the narrowly defined

BUILDING PERFORMANCE AND HUMAN REQUIREMENTS

Introductory remarks. In building science, the habitability of indoor environments is

The thermal environment

information on skin temperature. In general, thermoreceptors of the human skin give

Numerous actions may be initiated – consciously or unconsciously – by buildings'

A note on the information-related aspect of environmental relationships. While the

As mentioned earlier, one should be utterly careful with postulating strictly

BUILDING PERFORMANCE AND HUMAN IMPACT

cannot be evaluated without considering the behavior of people under emergency.

Empirical observations and models. There is a substantial and growing body of

and internal processes in inquiries may be unnecessary if not counterproductive.

ecology as a conceptual framework for our inquiry. In this framework, building

buildings.” Revisions to ASHRAE Standard 55, Energy and Buildings, 34:6

Proceedings of the 1992 Birmingham Symposium; Austrian and British

Occupant Behaviors and Energy Use:

Julia K. Day, Ph.D.1

Many high-performance building strategies, such as daylighting and natural

INTRODUCTION & BACKGROUND

Many high-performance building strategies offer a wealth of potential benefits

This study followed a sequential explanatory mixed-methods design. The

As previously discussed, this paper presents findings from two selected