close

Вход

Забыли?

вход по аккаунту

?

9780784481219.030

код для вставкиСкачать
International Conference on Sustainable Infrastructure 2017
Spatial and Geographic Patterns of Building Energy Performance:
A Cross-City Comparative Analysis of Large-Scale Data
Downloaded from ascelibrary.org by University Of Florida on 10/25/17. Copyright ASCE. For personal use only; all rights reserved.
Sokratis Papadopoulos1; Bartosz Bonczak2; and
Constantine E. Kontokosta, Ph.D., P.E., M.ASCE3
1
Ph.D. Candidate, Dept. of Civil and Urban Engineering, Center for Urban Science
and Progress, New York Univ., New York, NY 10003.
2
Associate Research Scientist, Center for Urban Science and Progress, New York
Univ., New York, NY 10003.
3
Assistant Professor, Dept. of Civil and Urban Engineering, Center for Urban Science
and Progress, New York Univ., New York, NY 10003. Email: ckontokosta@nyu.edu
Abstract
In recent years, there has been a growing interest in comparing and benchmarking the
energy use of buildings in cities. Existing frameworks such as the U.S.
Environmental Protection Agency (EPA) Energy star program and the U.S. Green
Building Council Leadership in Energy and Environmental Design (LEED)
certification provide coarse measures of energy performance and do not fully capture
specific features of the localized building stock that can influence consumption
patterns. Overall performance of existing buildings in a particular municipality can be
impacted by physical and occupancy characteristics, such as size, mass, age, and
implemented technologies, as well as the condition and extent of supporting
infrastructure and the morphology of the city. In addition, local land use and building
regulations, energy policies, and socio-cultural context can be significant drivers for
the adoption of energy efficient technologies and behavior. In this study, we analyze
publicly available energy disclosure data from five U.S. cities, collected through local
energy disclosure ordinances, in order to understand the spatial patterns of energy
consumption in varying urban environments. We use data on actual annual energy
consumption and physical and use characteristics for 2,250 office properties for the
year 2014. This robust dataset allows us to study differences between the building
stock in various municipalities, as well as to model direct building-to-building
benchmarking comparisons. The objective of this research is to develop the
foundation for a city-scale energy model that explains the impact of the urban built
environment on energy consumption between buildings of similar type, and to create
a robust, data-driven, and cross-city building energy efficiency benchmarking
framework.
© ASCE
International Conference on Sustainable Infrastructure 2017
336
International Conference on Sustainable Infrastructure 2017
Downloaded from ascelibrary.org by University Of Florida on 10/25/17. Copyright ASCE. For personal use only; all rights reserved.
INTRODUCTION/BACKGROUND
Increasing the energy efficiency of the urban building stock represents an
important step toward mitigating the risks of climate change and improving overall
well-being in cities. Furthermore, energy efficiency in existing buildings can lead to
significant cost and energy savings for owners, tenants, and the city as a whole (Asadi
et al. 2012). According to the U.S. Department of Energy (DOE), the building sector
accounts for 40% of greenhouse gas (GHG) emissions and primary energy use in the
U.S., suggesting that building energy use reductions can have a dramatic impact on
national and, consequently, global carbon emissions (Department of Energy 2011).
Cities and local governments have taken the lead in driving GHG reductions
and overall sustainability planning. Over the past several years, many municipalities
have implemented policies designed to reduce energy usage and emissions levels by
increasing efficiency and encouraging innovation. One of the most significant of
these policies is energy disclosure, also known as energy benchmarking. The first
jurisdictions to adopt mandatory benchmarking and disclosure laws were the District
of Columbia (2008) and New York City (2009) (Kontokosta 2013; Kontokosta et al.
2016). By the end of 2016, just eight years since the first law was passed, twenty-four
(24) municipalities have passed legislation requiring properties to regularly report
their energy and water consumption. Across cities, these policies differ by property
types (public, commercial, multifamily, residential) and building size thresholds
covered by the respective requirements. Also, the level of transparency varies by city,
from restricting data to internal administrative use to annual aggregated reports and
public access for each property record (Table 1) (Dillingham G. and BadoianKriticos M. 2016; Institute for Market Transformation 2016). This growing sample of
data gives researchers an unprecedented opportunity to examine factors that impact
building energy use in different markets and regions, and allows for benchmarking of
properties with their peers to evaluate relative energy performance (Kontokosta
2012). In the absence of such data, the market has relied on labeling or certification
programs (such as U.S. Green Building Council’s LEED rating system and
Environmental Protection Agency’s EPA Energy Star program) to assess whether a
building was more “sustainable” or energy efficient (Newsham et al. 2009). However,
the availability of building-level energy data enable more robust building- and cityscale analyses, which have the potential to accelerate the diffusion of energy efficient
technologies and practices (Kontokosta 2013).
In this study, we analyze publicly available energy disclosure data from five
U.S. cities, collected through local energy disclosure ordinances, in order to
understand the spatial patterns of energy consumption in varying urban environments.
We use data on actual annual energy consumption and physical and use
characteristics for 2,250 office properties for the year 2014. This robust dataset
allows us to study differences between the building stock in various municipalities, as
© ASCE
International Conference on Sustainable Infrastructure 2017
337
International Conference on Sustainable Infrastructure 2017
338
well as to model direct building-to-building benchmarking comparisons. The
objective of this research is to develop the foundation for a city-scale energy model
that explains the impact of the urban built environment on energy consumption
between buildings of similar type, and to create a robust, data-driven, and cross-city
building energy efficiency benchmarking framework.
Downloaded from ascelibrary.org by University Of Florida on 10/25/17. Copyright ASCE. For personal use only; all rights reserved.
DATA & METHODS
We use publicly available energy disclosure data obtained through the
respective cities’ official administration websites or open data portals to build a
repository of building-level energy use. These data are derived from the output of
ENERGY STAR Portfolio Manager®, an online tool for tracking the energy and
water consumption of buildings. As can be seen in Table 1, not all municipalities
require the same information to be disclosed. To maintain consistency among the
datasets, this analysis is focused on commercial office buildings. In order to
maximize the sample size, the analysis covers data reported for calendar year 2014,
which initially includes seven cities.
Table 1. Cities with adopted Building Energy Benchmarking and Disclosure Policies
(as of December 2016) (Institute for Market Transformation 2016).
City
Year
adopted
Public /
Government (ft2)
Commercial
(ft2)
Residential / Multifamily (ft2)
Public property
data
Washington, D. C.
2008
>= 10,000
>= 50,000
>= 50,000
Yes (2011-2015)
Austin, TX
2008
>= 10,000
>= 10,000
>= 5 units
/ all single-family
Yes (2013-2015
/ 2009-2013
single-family)
New York City, NY
2009
>= 10,000
>= 25,000
>= 25,000
Yes (2011-2014)
San Francisco, CA
2011
>= 10,000
>= 10,000
-
Yes (2011-2015)
Philadelphia, PA
2012
>= 50,000
>= 50,000
>= 50,000
Yes (2013-2014)
Seattle, WA
2012
>= 20,000
>= 20,000
>= 20,000
-
Boston, MA
2013
ALL
>= 35,000
>= 35,000 / >= 25 units
Yes ( 2015-2016)
Chicago, IL
2013
>= 50,000
>= 50,000
>= 50,000
Yes (2014-2015)
Denver, CO
2013
ALL
-
-
-
Minneapolis, MI
2013
>= 25,000
>= 50,000
-
Yes (2014-2015)
Cambridge, MA
2014
>= 10,000
>= 25,000
>= 50 units
Yes (2015)
Cook County, IL
2014
>= 35,000
-
-
-
Montgomery
County, MD
2014
ALL
>= 50,000
-
-
Atlanta, NC
2015
>= 10,000
>= 25,000
>= 25,000
-
© ASCE
International Conference on Sustainable Infrastructure 2017
Downloaded from ascelibrary.org by University Of Florida on 10/25/17. Copyright ASCE. For personal use only; all rights reserved.
International Conference on Sustainable Infrastructure 2017
339
Berkeley, CA
2015
ALL
ALL
ALL / (single -family
<= 4 units)
-
Boulder, CO
2015
>= 5,000
>= 20,000
-
Yes (2015)
Kansas City, MO
2015
>= 10,000
>= 50,000
>= 100,000
-
Portland, OR
2015
>= 20,000
>= 20,000
-
-
Salt Lake City, UT
2015
>= 3,000
-
-
-
Evanston, IL
2016
>= 10,000
>= 20,000
>= 20,000
-
Los Angeles, CA
2016
>= 7,500
>= 20,000
>= 20,000
-
Orlando, FL
2016
>= 10,000
>= 50,000
>= 50,000
-
Pittsburgh, PA
2016
ALL
>= 50,000
-
-
Portland, ME
2016
>= 5,000
>= 20,000
>= 50 units
-
Rockville, MD
2016
>= 50,000
>= 50,000
-
-
The primary variable of interest is the Weather Normalized Source Energy
Use Intensity (EUI) expressed in thousands of British Thermal Units per square foot
(kBtu/ft2). Energy use intensity is one of the most commonly used metrics for
describing building energy efficiency, and therefore allows for some degree of
uniformity across different geographies due to the weather normalization process
(ENERGY STAR 2017). Since the public data from Austin and Philadelphia do not
include this variable, they are excluded from the analysis.
Our energy disclosure data come from San Francisco, Chicago, Minneapolis,
Washington D.C., and New York City, representing 2,250 office properties. The
datasets include several energy-related variables in addition to EUI, such as total
annual energy consumption, ENERGY STAR Score, building age, size, location, and
other physical and spatial characteristics (City of New York 2014; City of Chicago
2014; City of San Francisco 2014; Washington D.C., Department of Energy &
Environment 2014; City of Minneapolis 2014). Due to the self-reported nature of the
disclosure data, it is essential to remove potential outliers or misreported entries.
Since the weather normalized source EUI distribution is positively skewed, we apply
a logarithmic transformation in the data to approximate a normal distribution.
Following, we remove all observations falling outside the threshold of two standard
deviations away from the mean (City of New York 2016; Kontokosta 2015). . As
shown in Table 2, the final integrated, cleaned dataset consists of 1,750 office
properties. Roughly 60% of the analyzed buildings are located in New York City,
which is home to the largest number of commercial buildings of any city in the U.S.
The smallest sample comes from Minneapolis with 74 observations.
© ASCE
International Conference on Sustainable Infrastructure 2017
International Conference on Sustainable Infrastructure 2017
340
Downloaded from ascelibrary.org by University Of Florida on 10/25/17. Copyright ASCE. For personal use only; all rights reserved.
Table 2. Building sample size before and after data cleaning
Dataset
No. properties before cleaning
No. properties after cleaning
Percentage
San Francisco
585
507
86.6
Chicago
153
139
90.8
Minneapolis
82
74
90.2
Washington, D.C.
529
445
84.1
New York City
1442
1085
75.2
Total
2791
1750
80.6
We begin our cross-city comparison of building energy performance with
basic descriptive statistics. The violin plots in Figure 1 show a combination of box
plot and a kernel density estimation of the underlying distribution for each individual
city. The dashed line represents the sample’s median, whereas the dotted lines
indicate the interquartile range. Although most cities’ median EUI is around 200
kBtu/ft2, San Francisco buildings consistently are found to be more efficient, with a
median EUI of approximately 130 kBtu/ft2. The majority of the violin plots shows a
positive skewness (i.e. longer upper tail) caused by higher intensity (and possibly less
efficient) commercial buildings.
Fig. 1. Violin plots showing distribution of EUI values for each city.
© ASCE
International Conference on Sustainable Infrastructure 2017
International Conference on Sustainable Infrastructure 2017
Downloaded from ascelibrary.org by University Of Florida on 10/25/17. Copyright ASCE. For personal use only; all rights reserved.
RESULTS & DISCUSSION
Each building observation is georeferenced based on the reported address
information using the Google Geocoding API in order to retrieve each property’s
specific location (Google Inc. 2017). The spatial distribution of benchmarked office
buildings for each city is represented in Fig.2. The primary concentration of office
properties tends to be located in and around central business districts with some
scattered observations in peripheral areas. To quantify the observed spatial patterns,
we estimate a Calculated Directional Distribution that shows geographical dispersion
and its direction based on the location of the properties. It is represented by Standard
Deviational Ellipse (Mitchell 2005). The size of an ellipse indicates the dispersion of
the audited properties. The ellipses shown in figure 2 are calculated for two standard
deviations which mean that roughly 95% of observations fall within their perimeter.
The black ones represent the dispersion based on the location of individual properties
and the red ones are weighted by total annual energy consumption. The difference in
the size and the location of those two indicates heterogeneity of energy demand
within the building sample. If both ellipses are aligned, as it is in Washington D.C., it
means that most of the audited buildings consume similar amount of energy. On the
other hand, smaller energy ellipses imply higher concentration of bigger
constructions or properties of a higher energy demand. The shift or rotation of this
ellipse in relation to building location ellipse indicates that energy hot spot is not
always fully aligned with buildings location.
Analyzing areas of the resultant ellipses with relation to total area of the city
within its administrative boundaries yields insights about general concentration of the
audited office properties. The highest concentration from building location
perspective can be observed in Chicago. At the same time Chicago characterizes with
highly skewed distribution resulting from properties located around O’Hare airport,
in the north-western edge of the city. The highest dispersion is observed in New York
City resulting from the multiple commercial business districts in the boroughs of
Manhattan, Brooklyn, and Queens. However analysis of energy demand dispersion
among office buildings indicates great share of Manhattan properties that
significantly overshadows outer boroughs. Similar pattern can be observed in San
Francisco, where most of office energy demand is concentrated in Financial District.
On the other hand, the study of the building location distribution in relation to energy
concentration proves that Washington D.C. demonstrates the most uniformly
distributed energy demand among analyzed cities. It can be inferred that majority of
the reporting properties are of similar size or characterize with close annual energy
consumptions levels.
© ASCE
International Conference on Sustainable Infrastructure 2017
341
Downloaded from ascelibrary.org by University Of Florida on 10/25/17. Copyright ASCE. For personal use only; all rights reserved.
International Conference on Sustainable Infrastructure 2017
Chicago, IL
342
Minneapolis, MN
New York City, NY
Washington, D.C.
San Francisco, CA
Fig. 2 Concentration of benchmarked office buildings within city limits with the
heatmap of property density, spatial dispersion measure (black ellipse) and spatial
dispersion measure weighted by total energy consumption (red ellipse).
Since one of the primary objectives of energy disclosure ordinances is to
catalyze the reduction of GHG emissions from existing buildings, it is important to
analyze the relationship between actual energy consumption and GHG emissions.
Figure 3 presents a double-axis chart showing the mean values of three variables of
interest: (i) weather normalized source EUI (down-pointing triangles), (ii) weather
normalized site EUI (up-pointing triangles), and (iii) GHG emissions intensity
(circles). Source EUI accounts for conversion efficiencies of the primary fuel sources
and transmission losses. Site EUI represents the end-use energy consumed at a
building, as is reflected in utility bills. San Francisco’s office building stock not only
exhibits lower EUIs, as discussed earlier, but also lower GHG emissions intensity,
with a mean of approximately 4 kgCO2/ft2 per year. Moreover, the divergence
between site and source EUI (vertical lines) provides important information about
energy source mix, including the adoption of renewable energy at scale. Another
interesting finding from Figure 3 is the relationship of energy and GHG intensity
across cities. Looking at Chicago, Minneapolis, Washington D.C., and New York
© ASCE
International Conference on Sustainable Infrastructure 2017
International Conference on Sustainable Infrastructure 2017
Downloaded from ascelibrary.org by University Of Florida on 10/25/17. Copyright ASCE. For personal use only; all rights reserved.
City, we observe similar values for mean source EUI (approximately 200 kBtu/ft2).
However, the GHG emission intensity pattern varies considerably, with New York
City emitting almost 50% less GHGs per square foot per year than Chicago. The
drivers of such divergence are an important area for further research.
Fig. 3. Chart showing energy and carbon intensity across cities.
Energy Star Score (ESS) is a regression-based performance metric for energy
efficiency in buildings (ENERGY STAR 2017). ESS has been recently criticized by
the scientific community due to its high uncertainty and model specification errors
(Kontokosta 2015; Scofield et al. 2014; Scofield 2014). According to Environmental
Protection Agency’s Portfolio Manager tool, buildings scoring above 75 (on a 1 to
100 scale) are eligible for ENERGY STAR certification. Figure 4 shows the fraction
of office buildings eligible for certification in each city. Interestingly, more than 65%
of Minneapolis office buildings are eligible for certification, although the city
exhibits the second highest GHG emission intensity from reported buildings. On the
other hand, New York City, which has the second lowest mean GHG emission
intensity, is found to have less than 50% of its office buildings eligible for
certification. Not surprisingly, almost 60% of San Francisco’s reported buildings
achieve high ESS.
© ASCE
International Conference on Sustainable Infrastructure 2017
343
Downloaded from ascelibrary.org by University Of Florida on 10/25/17. Copyright ASCE. For personal use only; all rights reserved.
International Conference on Sustainable Infrastructure 2017
Fig. 4. Bar chart of the fraction of reported buildings eligible for ENERGY STAR
certification.
Understanding the factors influencing building energy consumption, as well
as how energy is consumed by different building typologies is necessary for more
granular and targeted urban energy policy decision-making. Due to limitations in the
public data, only two factors, namely building age (Figure 5) and building floor area
(Figure 6), are discussed here. Figure 5 shows the relationship between building age
and EUI across the five cities. Given the age of New York City’s building stock, a
clear difference emerges across the analyzed cities. In general, a linear best-fit of the
data reveals a positive correlation between the year built and a building’s EUI for
New York City, Chicago, and Washington, DC. However, this finding does not
necessarily indicate that older buildings are more efficient. Factors such as higher
occupancy intensity and technology use in newer buildings may impact relative
energy performance in across building ages (Yohanis et al. 2008).
© ASCE
International Conference on Sustainable Infrastructure 2017
344
Downloaded from ascelibrary.org by University Of Florida on 10/25/17. Copyright ASCE. For personal use only; all rights reserved.
International Conference on Sustainable Infrastructure 2017
Fig. 5. Relationship between energy intensity and building age.
Similar to our finding for the relationship between EUI and building age, the
impact of building area on EUI is discussed in Figure 6. Unlike the results presented
above, building area does not appear to have a consistent effect on EUI. For instance,
New York City and San Francisco buildings show increasing EUI as size increases.
However, in Washington D.C. and Minneapolis the relationship is reversed, with EUI
slightly decreasing in larger buildings. There can be several arguments justifying both
trends. From one perspective, larger buildings are usually occupied by more energyintense tenants (e.g. large corporations with in-house servers, data centers etc.) that
would lead to higher whole-building energy use. On the other hand, larger buildings
can benefit from economies of scale and install more energy efficient equipment, as
well as have more sophisticated management systems. For instance, large buildings
are more likely to be equipped with more efficient, centralized heating and airconditioning systems or co-generation plants.
© ASCE
International Conference on Sustainable Infrastructure 2017
345
Downloaded from ascelibrary.org by University Of Florida on 10/25/17. Copyright ASCE. For personal use only; all rights reserved.
International Conference on Sustainable Infrastructure 2017
Fig. 6. Relationship between energy intensity and building size.
The findings discussed in this section reveal that buildings in U.S cities
exhibit both a range of energy performance levels, and different features influencing
that performance (i.e. building size, building age, disparity between source and site
EUI). Nevertheless, there are additional factors related to building energy
performance, such as occupancy intensity or end-use energy mix that should be added
in energy disclosure requirements and taken into account in the development of
comprehensive, city-specific benchmarking tools. Such tools facilitate the
establishment of robust energy performance metrics, with applications to city-wide
policy-making and peer-to-peer energy performance comparison.
CONCLUSION & FUTURE WORK
This research is a preliminary attempt to utilize the growing availability of
energy disclosure data to understand patterns of energy use in existing buildings
across multiple cities. For the five cities studied, we observe non-trivial differences in
spatial dispersion of covered buildings, notable variations in EUI and GHG
intensities, and divergent effects of building age and size on energy use.
As more cities begin to mandate energy data reporting, this study provides the
foundation for a deeper investigation of energy use patterns across cities and urban
© ASCE
International Conference on Sustainable Infrastructure 2017
346
International Conference on Sustainable Infrastructure 2017
Downloaded from ascelibrary.org by University Of Florida on 10/25/17. Copyright ASCE. For personal use only; all rights reserved.
environments. Future work will include the development of an integrated statistical
model that captures the energy dynamics of each city and the extension of this
analysis to other building typologies, such as multifamily housing. Finally, we will
integrate non-energy related data sources, such as land use and demographics, to
derive further insights into the causes of varying spatial patterns of consumption at
the city scale.
REFERENCES
Asadi, E., Da Silva, M. G., Antunes, C. H., & Dias, L. (2012). Multi-objective
optimization for building retrofit strategies: a model and an
application. Energy and Buildings, 44, 81-87.
City of Chicago (2014). “Chicago energy benchmarking”,
<https://data.cityofchicago.org/Environment-SustainableDevelopment/Chicago- Energy-Benchmarking/xq83-jr8c>.
City of Minneapolis (2014). “2014 energy benchmarking report”,
<http://www.ci.minneapolis.mn.us/www/groups/public/@health/documents/i
mages/wcmsp- 183032.pdf>.
City of New York (2014). “Energy and water data disclosure for Local Law 84”,
<https://data.cityofnewyork.us/Environment/Energy-and-Water-DataDisclosure-for-Local-Law-84-/rgfe-8y2z>.
City of San Francisco (2014). “Existing Commercial Buildings Energy Performance
Ordinance Report”, <https://data.sfgov.org/Energy-and-Environment
/Existing- Commercial-Buildings-Energy-Performance-O/j2j3-acqj>.
Department of Energy (2011). “Buildings energy databook.” Energy Efficiency &
Renewable Energy Department.
Dillingham G. and Badoian-Kriticos M. (2016). “Adoption of benchmarking and
transparency policies in the united states.” Proceedings of the 2016 ACEEE
Summer Study on Energy Efficiency in Buildings. American Council for an
Energy-Efficient Economy, ACEE.
ENERGY STAR (2017). “Energy star portfolio manager website”,
<https://www.energystar.gov/buildings/facility-owners-andmanagers/existing- buildings/use-portfolio-manager>.
Google Inc. (2017). “Google maps API: Geocoding API,
<https://developers.google.com/maps/documentation/geocoding/start>.
Institute for Market Transformation (2016). “Building rating - sharing transparency
for a more efficient future”, <http://www.buildingrating.org>.
Kontokosta, C. E., Bonczak, B., and Duer-Balkind, M. (2016). “DataIQ - a machine
learning approach to anomaly detection for energy performance data quality
and reliability.” Proceedings of the 2016 ACEEE Summer Study on Energy
Efficiency in Buildings. American Council for an Energy-Efficient Economy.
© ASCE
International Conference on Sustainable Infrastructure 2017
347
International Conference on Sustainable Infrastructure 2017
Downloaded from ascelibrary.org by University Of Florida on 10/25/17. Copyright ASCE. For personal use only; all rights reserved.
Kontokosta, C. E. (2015). “A market-specific methodology for a commercial building
energy performance index.” The Journal of Real Estate Finance and
Economics, 51(2), 288–316.
Kontokosta, C. E. (2013). “Energy disclosure, market behavior, and the building data
ecosystem.” Annals of The New York Academy of Sciences, 1295(2013), 34–
43.
Kontokosta, C. (2012). Local Law 84 Energy Benchmarking Data: Report to the New
York City Mayor’s Office of Long-Term Planning and Sustainability.
Newsham, G. R., Mancini, S., & Birt, B. J. (2009). Do LEED-certified buildings save
energy? Yes, but…. Energy and Buildings, 41(8), 897-905.
Mitchell, A. (2005). The ESRI Guide to GIS Analysis: Spatial Measurements &
Statistics. Vol. 2. ESRI Press
Scofield, J. (2014). “U.S. EPA Energy Star benchmarking scores for medical office
buildings lack scientific validity.” manuscript in preparation. http://www.
oberlin.edu/physics/Scofield/pdf_files/Scofield_2014_medical_office. pdf.
Scofield, J. H. et al. (2014). “Energy star building benchmarking scores: Good idea,
bad science.” Oberlin College study for American Council for an Energy
Efficient Economy (ACEEE).
Washington D.C., Department of Energy & Environment (2014). “2014 Building
Benchmarking Dataset, <https://doee.dc.gov/node/1150817>.
Yohanis, Y. G., Mondol, J. D., Wright, A., & Norton, B. (2008). Real-life energy use
in the UK: How occupancy and dwelling characteristics affect domestic
electricity use. Energy and Buildings, 40(6), 1053-1059.
© ASCE
International Conference on Sustainable Infrastructure 2017
348
Документ
Категория
Без категории
Просмотров
3
Размер файла
884 Кб
Теги
9780784481219, 030
1/--страниц
Пожаловаться на содержимое документа