article review paper

Journal of Building Engineering 42 (2021) 102475

Available online 18 April 2021 2352-7102/© 2021 Elsevier Ltd. All rights reserved.

2019 energy benchmarking data for LEED-certified buildings in Washington, D.C.: Simulation and reality

Ming Hu University of Maryland, School of Architecture, Planning, Preservation, USA

A R T I C L E I N F O

Keywords: Energy benchmarking For LEED-Certified buildings Simulation Reality

A B S T R A C T

This study aims to understand the actual performance difference between LEED buildings and non-LEED buildings. Since 2012, the District of Columbia (DC) has amended regulations so that all buildings must report their building energy use. We have cross-referenced the most recently published data of the 2019 DC energy benchmarking database with the U.S. Green Building Council’s LEED project database to identify DC properties in both databases that are expected to reduce building operating energy use and greenhouse gas emissions. We compared LEED office buildings and non-LEED-certified office buildings using their reported operating source and site energy use intensity (EUI). The results show that LEED office buildings do not perform better at any of the certified levels. On the contrary, those reported LEED buildings collectively use 17% more source energy and 13% more site energy than non-LEED buildings. Among the different LEED levels, LEED Silver appears to perform slightly better than the other LEED levels. Meanwhile only around 33% of qualified LEED office buildings reported their actual energy use according to the DC regulation. The purpose of this study is not to criticize the LEED rating system; instead, we want to improve the system in order to meet DC’s carbon neutrality goal. To this extent, we conclude that the U.S. LEED rating system can benefit from learning from other green building rating systems that include reporting and verification as prerequisite requirements.

1. Introduction

Commercial buildings are a large energy consumer. In 2012, large commercial buildings in the United States had more than 25,000 ft2 of floor space each, accounting for about 69% of total commercial building energy use [1]. Larger commercial buildings, covering over 200,000 ft2, only accounted for 1% of total commercial buildings, but they contrib- uted to 26% of the total commercial building energy consumption [1]. Among commercial buildings, offices accounted for 18% of total floor space and 20% of overall site energy use, with the latter being the largest category (for energy-consuming) in commercial buildings [1]. Reducing office energy use not only has a bigger impact on overall commercial building energy performance improvement but also can provide some valuable lessons for other commercial building categories.

Leadership in Energy and Environmental Design (LEED), by the U.S. Green Building Council (USGBC), is the most widely used green building certification in the United States. Since its establishment in 2000, the credibility of LEED certification has been debated. Most criticism is centered around the actual energy performance of LEED-certified buildings and whether LEED actually implies energy efficiency and

sustainability [2]. Amiril et al. (2019) reviewed 44 peered-reviewed articles focusing on the energy performance of LEED buildings. Ten papers concluded that LEED buildings had higher energy efficiency, eight studies stated the opposite conclusion, and the remaining papers did not articulate the comparison. However, the consensus is that the energy efficiency of LEED buildings is questionable, and modifications to the LEED Energy and Atmosphere category are recommended to improve actual building performance [2]. In the past several years, the mid-Atlantic region has remained the “LEED building leader,” which is perceived as a progressive region leading the efforts to pursue sustain- able design [3]. The District of Columbia has consistently led the nation in registering and certifying LEED buildings. In 2018, the city certified 61.74 ft2 of space per resident across 145 building projects, which is more than 10 times the per capita number for top-ranked states, such as Illinois [3]. It is necessary to investigate the actual performance of those LEED buildings in DC.

In 2008, DC passed the Clean and Affordable Energy Act (CAEA), which requires that all buildings with a gross floor area of 50,000 ft2

(4645 m2) or greater to report their actual building energy and water use annually. The DC government must also annually benchmark and

E-mail address: [email protected].

Contents lists available at ScienceDirect

Journal of Building Engineering

journal homepage: http://www.elsevier.com/locate/jobe

https://doi.org/10.1016/j.jobe.2021.102475 Received 20 September 2020; Received in revised form 25 March 2021; Accepted 26 March 2021

Journal of Building Engineering 42 (2021) 102475

2

disclose the energy and water efficiency of district government buildings over 10,000 ft2 (929 m2) [4]. The benchmarking is done according to the ENERGY STAR Portfolio Manager® by the U.S. Environmental Protec- tion Agency (EPA), which is an industry-standard free online tool [5]. It was developed to provide a method for comparing the energy con- sumption of a commercial building with that of similar activities, adjusting for size, climate, and operational characters [6,7]. This method makes it possible to determine the actual LEED building per- formance by cross-referencing DC energy benchmarking data (from the DC government website) and LEED project inventory (from the USGBC website) to understand whether LEED office buildings perform better than non-LEED buildings.

2. Materials and method

The overall data cleaning and searching process is illustrated in Fig. 1.

2.1. Energy benchmarking data

Since 2012, the District of Columbia has released benchmarking data for more than a thousand buildings under the benchmarking law [8]. It includes all private buildings over 50,000 gross ft2 within the District of Columbia, including multifamily residences, offices, education build- ings, mixed-use buildings, hospitals, libraries, hotels, K-12 schools, su- permarkets, colleges/universities, restaurants, and police stations, among others. The following data was reported: 1591 buildings in 2013, 1849 buildings in 2014, 2015 buildings in 2016, 1847 buildings in 2017, 2095 buildings in 2018, and 2162 buildings in 2019. For this study, we use the dataset from 2019 (based on 2018 operations), which has the highest data reporting compliance: 1343 buildings. Among all buildings included in the 2019 report, we excluded buildings that were exempt from 2019’s disclosure, those that currently have data under review, and those with no report received, which resulted in 1333 buildings. Among those buildings, the building types included commercial office buildings (478), K-12 schools (29), and multifamily housing (556); the remaining buildings cover 35 other building types. The released data includes both descriptive and energy use information; Table 1 Table 1 lists the specific information released for each building [9].

The USGBC publishes a database of all LEED-registered buildings worldwide. When we extracted the inventory from the USGBC website (May 2020), USGBC had nearly 126,177 buildings; 97,378 in the United States, with 2917 in Washington, D.C., and none of the DC buildings were listed as confidential without further identifying information. Among 2917 buildings, 1530 of those buildings were certified at different levels, and only 7 buildings were reported to have been certi- fied before January 1, 2008, the beginning of the year for which DC energy benchmarking data was mandated.

Another database we used as a baseline comparison was the Com- mercial Buildings Energy Consumption Survey (CBECS) data, which was published on the U.S. Energy Information Administration (EIA) website. EIA is the organization responsible for conducting surveys of building energy use and managing and publishing CBECS data. In 2018, EIA

released the most recent CBECS data, which was based on a 2015 survey. CBECS provided a snapshot of the U.S. commercial building sector characteristics and energy performance [10].

2.2. Energy measurement metric: gross EUI

The energy use intensity was chosen as a measuring unit and is measured by the energy use kWh (kBtu in the U.S.) divided by the total gross floor area of a building in m2 (ft2 in the U.S.). Regarding individual buildings’ energy efficiency, both the mean EUI [11] and median EUI [12] have been used to compare LEED buildings and non-LEED buildings in previous studies [8]. The EPA has been using the mean EUI as a benchmark tool to measure the decrease of energy use in U.S. buildings [13]. However, simply using the reported EUI of individual buildings to calculate the mean or median EUI can lead to misleading results. This study adopted a method proposed and verified by Dr. John H. Scofield in 2013 [14]. He demonstrated that using the EUI (mean or median) of an individual building for calculation of the mean or median EUI of a set of buildings actually treats small and large buildings equally, despite large buildings contributing more total energy. This discovery has been clearly demonstrated in his Chicago LEED building study [15]. He showed that even when two buildings’ sets have identical numbers of buildings and an identical total floor area, it is possible for one set of buildings to use more energy than the other set, despite having a lower mean or median EUI [16]. Therefore, Scofield proposed a mathemati- cally consistent and useful generalization of EUI for a set of N buildings, where the gross EUI (GEUI) is calculated (see the following equation), which is the mathematical equivalent of the area-weighted mean EUI, e[16].

e = ∑

Ej ∑

Aj

where ∑

Aj is the total floor area for the building set, and ∑

Ej is the total energy use for the building set.

GEUI is used by the U.S. Energy Information Administration for the U.S. Commercial Buildings Energy Consumption Survey, which has been used by many researchers to study building energy use and efficiency. GEUI can be used to compare sets of buildings with different numbers of

Fig. 1. Data-acquiring process.

Table 1 Building properties extracted from the Washington, D.C. public Building Benchmarking Portal [9].

Property Description

Energy Use CO2 Emissions

Property name Energy star score Total emissions (GHG) Address Source EUI Total emissions intensity

(GHG) Zip code Site EUI Property type Weather normalized source

EUI Owner Weather normalized site EUI Ward Electricity use Year built Natural gas use Gross floor area

M. Hu

Journal of Building Engineering 42 (2021) 102475

3

buildings and different total floor areas. In addition, source energy is used in this study, since source energy provides a more accurate mea- surement of understanding the total energy impact of a building.

2.3. LEED building identification

DC benchmarking data does not include LEED certification infor- mation to identify the LEED-certified buildings that also reported the actual building energy use in the DC benchmarking database. We cross- referenced the 2019 benchmarking database with USGBC’s LEED project directory, extracted in May 2020, which included projects registered by December 2019. There are 643 projects with a floor area over 50,000 ft2

(4650 m2) that were certified at levels from Certified to Platinum. The rating system includes New Construction, Core and Shell, School, BD + C, and EB O + M [17].

Then, based on the project names and street addresses, we searched for all of the LEED-certified buildings in the DC benchmarking data file. This process allowed us to identify 205 office buildings that were included in both the DC benchmarking database and LEED projects li- brary. Among the 205 buildings, there are five buildings that share the same street address, which means that one property corresponds with two or more different LEED projects at different certification levels. For example, one LEED project was certified as LEED Silver and another was certified as LEED Gold. However, in the benchmark report, there is only one project reported. The reported floor areas listed in the two databases were not identical; therefore, it was hard to determine which LEED project was reported in the benchmarking database. Consequently, we excluded those projects in the analysis. We also matched the gross floor area in the two databases; not all buildings have a matching reported floor size. We excluded the projects that have a difference in the re- ported floor area that is over 10%. There are a couple of buildings that have been certified twice; for instance, one building was certified as Silver in 2010 and then certified as Platinum in 2018. In this study, the higher and more recent certification was used to reflect a recent energy performance measurement. This resulted in 197 buildings for the anal- ysis; the breakdown of the 197 office buildings is illustrated in Table 2.

3. Results

In 2019, according to the DC benchmarking database, as mentioned earlier, 1333 buildings are in compliance with the reporting re- quirements. Among the 1333 buildings, there is a total of 478 office buildings, which account for 36% of all buildings in the benchmarking database. The largest category in the benchmarking database is multi- family buildings (556); although this study focuses on office buildings, future studies will look into multifamily housing. The cross-referencing with the LEED project inventory resulted in the identification of 197 LEED-certified office buildings in both databases. The number of other LEED-certified building types did not have a sufficient match in the benchmarking database. The basic statistics for LEED-certified and non- LEED (other) buildings are summarized in Table 3; the data includes the building year (median), total floor area, gross site and source EUI, and CO2 intensities. The value of the relative standard error (RSE) was extracted from the U.S. Energy Information Administration’s website. According to EIA, the relative standard error is a measure of the

reliability of precision in the survey statistics. The value for the relative standard error can be used to construct confidence intervals and to perform hypothesis tests by standard statistical methods [18].

3.1. DC buildings vs. national and regional buildings

It is important to compare the energy use pattern between DC buildings and national and regional office buildings. Table 4 lists the number of buildings, total gross floor area, gross site EUI, gross source EUI, and associated greenhouse gas (GHG) emissions intensity. The GHG intensity is calculated using fuel data combined with e-grid information about the regional electric grid [29]. The national reported median gross EUI for office buildings was reported by the U.S. ENERGY STAR Port- folio Manager in 2018; those EUI figures were based on 2018 CBECS survey data. Office buildings nationwide have a gross site and source EUI of 52.9 kBtu/ft2 (166.9 kWh/m2) and 116.4 kBtu/ft2 (367.2 kWh/m2), respectively. Comparing these national values, in Washing- ton, D.C., office buildings use considerably more energy, both site and source energy.

We then dove deeper to understand which office building group/set had the highest energy use, and we found two important indicators of energy performance: building size and building age. Table 5 lists the properties from several U.S. building sets extracted from the 2018 CBECS survey data. These include all U.S. office buildings (CBECS-Off), large (>4645 m2 or 50,000 ft2) U S. office buildings (CBECS-larg-Off), all regional buildings (CBECS-reg), regional office buildings (CBECS-Reg- Off), and large regional office buildings (CBECS-Reg-larg-Off). Regional buildings are defined as being in the Mid-Atlantic census region and climate zones 4a and 4b. The CBECS data shows that office buildings tend to have higher site and source EUI than other buildings in the Mid- Atlantic region, at±35.7%. Larger office buildings have a slightly higher energy intensity compared to smaller ones, at±11.5%. It is important to note that regional (Mid-Atlantic) office building sets have higher site and source EUI than those of national building sets. DC office building sets have higher energy use intensity compared to the national median but lower energy use intensity than that of regional building sets.

As for the age of buildings, Fig. 2(a) shows that DC LEED office

Table 2 LEED office buildings included in the study.

LEED Level

Buildings (N)

Medium Gross Floor Space

GEUI (source)

GEUI (site)

(m2) kWh/m2 kWh/m2

Certified 16 9073 487.1 155.2 Silver 54 21,788 448.6 171 Gold 104 26,219 490.9 178.2 Platinum 24 22,977 469.1 183.9

Table 3 Summary of LEED and non-LEED office buildings in the DC benchmarking database. Median year of construction, total floor area.

Building Subset

N Median Year

Floor Gross Area

Gross Site EUI Gross Source EUI

(m2) kwh/ m2

RES kWh/ m2

RES

DC LEED Offices

197 1988 25,163 186.4 2.7% 501.3 2.7%

DC non- LEED Offices

284 1980 15,147 151.1 2.1% 393.1 2.1%

Table 4 Summary of reported buildings in DC benchmarking database of office buildings and national office buildings extracted from CBECS 2018 survey data. The gross mean EUI is calculated using area weighting, and relative standard errors represent standard deviation.

Building Type

N Floor Gross Area

Building Age

Gross Site EUI

Gross Source EUI

CO2 Intensity

(m2) (kWh/ m2)

(kWh/ m2)

DC Offices

481 19242 1982 184.2 488.9 5.7

National Offices

1,012,000 1487 1975 166.9 367.2 5.4

M. Hu

Journal of Building Engineering 42 (2021) 102475

4

buildings are newer than the average DC office buildings and national office buildings; 47% of DC LEED buildings were built after 1990 (new or renovated), whereas nationwide, only 30% of office buildings were built after 1990. It has been shown in CBECS 2003 and 2012 data that newer office buildings tend to have higher source EUI than older office buildings [20]. Scofield also found from the NYC benchmarking report that older office buildings had lower site EUI than newer office buildings [Scofield 2013]. Fig. 2(b) demonstrates that DC LEED offices have a larger building size; in this study, buildings with a floor area over 50, 000 ft2 were defined as larger buildings. Additionally, 95% of DC LEED office buildings are large buildings, while only 84% of office buildings nationwide are described as large buildings. Based on 2018 CBECS data, newer buildings tend to be larger than older buildings [21]. The two building characteristics when combined, a larger floor space and newer construction, contribute to the higher EUI of DC LEED office buildings, both in source and site energy.

3.2. LEED vs. non-LEED

We first looked at 197 office buildings. In general, LEED office buildings do not exhibit reduced on-site or source energy consumption. Fig. 3(a) represents a comparison of the distribution of source EUI for LEED buildings (in green) and non-LEED buildings (in orange). On the X- axis, the first bin represents the buildings with 0 < EUI≤20kBtu/ft2, and

the last bin corresponds to source EUI≤360kBtu/ft2. Figure 3(a) shows an area-weighted histogram in which each bar represents the percentage of the total building gross floor area having site EUI that falls within a particular bin. For example, the first bin represents a floor area with 0 ≤ EUI ​ ≤ 20kBtu/ft2 while the 20th bin represents the range 380 ≤ EUI ​ ≤ 360kBtu/ft2. The Y-axis indicates the value percent of the building floor areas that fall into a certain EUI bin. The green bars above the horizontal line present the area-weighted source EUI of LEED buildings (n = 197), and the orange bars below the horizontal line are the area-weighted source EUI of non-LEED buildings (n = 284). The graph demonstrates that the mean source GEUI for LEED offices is higher than that of non-LEED office buildings. Next, we further calculated the area-weighted means, which showed the area-weighted mean of source GEUI of LEED buildings as 17.13% higher than that of non-LEED office buildings. In order to verify the statistical significance of the two sets of different mean EUIs, the standard student t-test with two samples of equal variance [30] was performed, which produced a t-value of 4.65, corresponding to a two-sided p-value of 0.0000043. This means that there is more than a 99.9% chance that this difference is not accidental. Thus we concluded that in 2019, DC LEED offices, on average, used 17.13% more source energy per unit area than did non-LEED offices. The same process has been repeated for site energy use, with the results shown in Fig. 3(b). DC LEED offices, on average, used 13.23% more site energy per unit area than did DC non-LEED office buildings.

Table 5 Breakdown and characteristics of LEED buildings.

Building Sets N Gross Floor Area (m2) Site GEUI (kwh/m2) Source EUI (kwh/m2)

Building weight Area weight Building weight Area weight

DC-LEED-off 197 25,163 612 671 1686 1804 CBECS-off 1013 1,487,098,961 600 662 1322 1340 CBECS-larg-off 53 794,320 698 702 1782 1800 CBECS-reg 504 1,044,415,975 617 603 1128 1102 CBECS-reg-Off 94 275,457,513 837 883 1751 1767 CBECS-reg-larg-off 56 69,677 933 961 1832 19 1882

Fig. 2. Year of construction and buildings’ size comparison.

Fig. 3. Source and site EUI comparison of DC LEED and non-LEED buildings.

M. Hu

Journal of Building Engineering 42 (2021) 102475

5

t-Test was conducted to verify whether there is difference between LEED building energy use and those of non-LEED building, both site EUI and source EUI were studied. The results from Table 6 confirmed there is energy use intensity difference between LEED and Non-LEED. (P < 0.05).

3.3. The performance of different LEED levels for buildings

The 197 LEED office buildings can be broken down into different certification levels: Certified, Silver, Gold, and Platinum. The break- down and characteristics of these LEED buildings are listed in Table 7.

Fig. 4 shows that LEED Gold buildings account for more than 50% of LEED buildings, followed by LEED Silver.

Fig. 5 shows that when comparing site energy use, LEED buildings do not perform better than non-LEED buildings, and this is true at all LEED levels. Compared to non-LEED office buildings, LEED Certified buildings use 3% more gross site energy, LEED Silver buildings use 13% more, LEED Gold buildings use 18% more, and LEED Platinum buildings use 22% more. Additionally, in contrast to non-LEED buildings, LEED Certified buildings use 24% more gross source energy, LEED Silver buildings use 14%, LEED Gold buildings use 24%, and LEED Platinum buildings use 19% more. The results for LEED Gold and LEED Silver buildings are significant, at the 95% confidence level, whereas the LEED Certified and LEED Platinum results are less so (85% confidence level).

Among the LEED categories, LEED Platinum and LEED Certified per- formed worse than LEED Silver and LEED Gold. LEED Silver was the best- performing category among all LEED buildings, for both source and site EUI. In terms of performance distribution, for gross source energy, LEED Gold had a much higher concentration in the GEUI range, between 120 and 180, and other LEED categories had similar concentrations in the GEUI range, between 120 and 220. For gross site energy, LEED Silver buildings had a slightly higher concentration in the GEUI range, be- tween 50 and 60, and other LEED buildings had a normal distribution, between 20 and 80.

4. Discussion

4.1. How this study compares to previous results

There are limited studies on the actual performance of sustainable buildings, and very few on LEED buildings due to unavailable perfor- mance data. To the author’s knowledge, there are two studies that are similar to this project, which can be compared to the results of this study. Scofield (2013) studied 953 large New York City office buildings from the 2011 ENERGY STAR Energy Performance Rating database and found 21 matching LEED buildings. Regarding energy consumption and greenhouse gas emissions, he concluded that LEED buildings do not perform better when compared with non-LEED buildings. Moreover, he found the LEED Gold level outperformed other buildings by 20% while LEED Silver and LEED Certified office buildings underperformed compared to other office buildings [22]. Saldanha et al. (2016) cross-referenced LEED data with a New York City local law (LL84) that requires both city-owned and private buildings to report their actual annual energy and water use, and they identified 91 LEED buildings, including 66 offices and 25 multifamily buildings. The analysis results

confirmed Scofield’s study: they found NYC LEED office buildings per- formed slightly worse than non-LEED buildings, with an average source EUI that was about 7% higher. Similarly, LEED multifamily buildings performed worse, with an approximate average source EUI 30% over that of non-LEED buildings [23]. Later, in 2018, Scofield compared Chicago LEED buildings against Chicago benchmarking data, and found that for offices, K-12 schools, and multifamily buildings, LEED buildings did not use less source energy than similar non-LEED buildings. On the contrary, LEED schools used 17% more source energy than did other non-LEED schools [24]. The findings from our study on the District of Columbia office building performance are aligned with those of previous studies. A significant energy performance gap exists for LEED buildings across different regions and building types. An energy performance gap is the difference between the project (simulated) energy performance and the actual performance delivered in operation. This gap has been a known problem for a while [25]. Cali et al. (2016) studied refurbished German houses and found that the missed predicted saving varies be- tween 41% and 117% [26]. In the UK, a reported by the Building Per- formance Evaluation programme looked at 50 modern and green buildings and found that non-residential buildings were not meeting performance expectations. The reported showed supermarkets, offices, schools and health centers were normally using up to 3.4 times more energy than they were designed for and produced on average 3.8 times the predicated carbon emissions [27].

Experts in the building industry often relate this gap to four potential causes: flaws in the construction caused by builders, overly complicated

Table 6 t-Test results: comparision of energy use tensity between LEED building and Non-LEED.

Comparison site energy use source energy use

df 474 474 t Stat 4.336410667 4.651497681 P(T ≤ t) one-tail 8.85173E-06 2.14028E-06 t Critical one-tail 1.64807466 1.64807466 P(T ≤ t) two-tail 0.000017703 0.000004281 t Critical two-tail 1.964981363 1.964981363

Table 7 Breakdown of DC LEED building characteristics.

Office Building Sets

N Floor Area (total) ft2

Site EUI (kBtu/ft2)

Area weight

Source EUI (kBtu/ft2)

Area weight

Building weight

Building weight

LEED (Cert)

16 2,731,622 48.1 49.2 147.1 154.4

LEED (Silver)

55 15,215,805 53.7 54.2 148.7 150.7

LEED (Gold)

101 34,681,887 56.1 56.6 151.2 154.5

LEED (Plat)

23 5,671,755 52.8 58.3 148.7 149.6

DC- non- LEED

284 58,357,176 63.7 64 162.5 168.2

Fig. 4. DC LEED building breakdown.

M. Hu

Journal of Building Engineering 42 (2021) 102475

6

energy-saving technologies [],unpredicted occupant behavior [Cali et al., 2016], and inaccurate energy modeling [28]. Recently, the focus has been on occupants’ behavior and the reliability of energy modeling. Studies have been conducted to determine the factors causing the in- accuracy of the energy modeling. Knowledge of and experience with the energy model play an important role. It was found that modelers often underestimate operating hours, which could possibly be due to opti- mism bias or lack of experience in accounting for uncertainty [U.K Government]. To date, the large gap between predicted energy saving and actual performance has not been sufficiently addressed even after recognition of the issue. There is high potential harm in keeping such an issued unaddressed for long. If the assumption continues that all LEED buildings can perform at the level they are initially projected at, this misperception may lead to LEED buildings actually using more energy and emitting more carbon.

Unlike the study of NYC LEED buildings, in our DC study, LEED Gold office buildings do not perform better than non-LEED buildings. Instead, they have about 24% higher source energy consumption and 18% higher site energy than those of non-LEED office buildings. LEED-certified buildings use 24% higher source energy and 3% higher site energy than those of non-LEED buildings. LEED Silver buildings use 14% more source energy and 13% more site energy. LEED Platinum buildings use 19% more source energy and 22% more site energy. Collectively, LEED office buildings, at all levels of certification, consume 17% more source energy and 13% more site energy compared to other non- LEED office buildings in the district (refer to Table 8).

There are two potential explanations for the higher energy use. The first is regarding the buildings’ operation. The operating hours can be longer in LEED buildings than those of non-LEED buildings, and the plug-load can be higher in LEED buildings than that in non-LEED buildings. The target or designed EUIs of those LEED buildings are not available on USGBC’s website; therefore, it is impossible to conclude whether those buildings perform better or worse than they are designed for, which makes it even more difficult to understand what energy modeling parameters the design teams used in order to achieve LEED certification. The second cause for the higher energy use intensity could be directly contributed by the building system operations. If the actual operations of the buildings are the same as those modeled or predicted and the actual EUIs are in fact higher than the predicted EUIs, then the design building system is most likely not installed or operating as

designed for or the equipment and system have malfunctioned. These types of problems have indeed happened previously.

Table 7 also illustrates that the results from measuring the source and site energy use are significantly different. For example, LEED Certified, LEED Silver, and LEED Gold buildings have higher EUIs compared to source energy. When only comparing site energy, LEED Certified build- ings perform at almost the same level as non-LEED buildings, at only 3% higher. The site energy performance and saving do not directly translate into source energy use and saving due to the energy source differences. The U.S. EPA defines building source energy as including off-site energy losses [27]. In the United States, the primary energy used to produce and transmit electricity on average is more than three times that of the actual electricity delivered to the buildings [1]. In this sense, using source energy for building performance provides a more accurate assessment of building performance. The U.S. national average source-to-site energy conversion factor for electricity is 2.18; for natural gas it is 1.05 [29]. Office buildings in the United States typically use a mix of fuels (natural gas and electricity), and newer buildings tend to have all-electricity use since moving towards 100% renewable electricity has been a global trend for the past five years [30]. This might be able to explain why the ratio between source and site energy at each LEED level is different. The overall weighted source and site EUIs are dependent on the mix within each building and the ratio of those buildings in corresponding levels.

Another important fact worth mentioning is the missing reported data. In the District of Columbia, there is a total of 643 LEED office buildings larger than 50,000 ft2 that were certified by the end of 2019. Only 30% of those buildings reported their actual energy use in the year 2019; if the reported floor space is included, only 33% reported the actual performance data. Among the 643 buildings, 72 buildings received their certification in the year 2019, so they do not have enough operating annual data to report; therefore, they can be exempted. However, there are still 571 buildings that need to report their operating data. As mentioned at the beginning, all privately owned commercial buildings over 50,000 gross ft2 are required to measure and report their energy use. Failure to do so can result in fines of up to $100 per day for non-compliance. According to the data, non-significant incompliance stands at 66.6%, which is surprising and, at the same time, alarming. It indicates that more stringent benchmarking rules need to be put in place, and a more effective monitoring system needs to be implemented to make sure all buildings meet the requirements and actually report

Fig. 5. Comparison of LEED and non-LEED buildings (GEUI site and GEUI source).

Table 8 Gross EUI comparison between non-LEED buildings and LEED buildings at different levels.

Number SITE GEUI SOURCE GEUI High(%) SITE GEUI High(%) SOURCE GEUI

NON-LEED 284 47.9 124.6 LEED(all) 197 54.3 145.9 13% 17% LEED(Certified) 16 49.2 154.4 3% 24% LEED(Silver) 55 54.2 142.4 13% 14% LEED(Gold) 101 56.6 154.5 18% 24% LEED(Platinum) 23 58.3 148.7 22% 19%

M. Hu

Journal of Building Engineering 42 (2021) 102475

7

their operating data. Meanwhile, this also brings to attention a very important concern of the LEED building market: Why did those LEED buildings not report their actual energy operating data? Even some Platinum buildings did not report their data. If we cannot hold the green building rating system accountable for its actual perfor- mance, how can we be certain that “green rated” buildings can help to reduce energy and carbon emissions? How can we be sure those green buildings do not harm our environment?

4.2. Limitations of this study

The primary limitation of this study is directly related to the insuf- ficient reporting of LEED office buildings in DC. As mentioned in the previous section, around 67% of LEED buildings did not report their actual performance data, thus the assumptions and conclusions made are based on 33% of LEED office buildings in DC, which may have skewed the analysis and results. The second limitation is rooted in LEED being a voluntary rating system opposed to being an energy or building code, so the sample of LEED buildings may be biased as compared to the DC benchmarking of buildings, many of which were built under mandated energy codes. The above two limitations can be mitigated by examining a larger data set, either of LEED buildings in other regions or by mandating more DC LEED office buildings to comply with the DC energy benchmarking rule.

4.3. Summary and next steps

The purpose of this study is not to criticize the LEED rating system but to instead encourage improvements to the system in order to meet DC’s carbon neutrality goal. The LEED rating system, together with other green building rating systems, has proved to be a powerful market driver for sustainable design movement. There are many studies claiming that LEED buildings can contribute to an energy use reduction [31], a carbon emissions reduction [32], and greater human health benefits [33]. These all hinge on whether LEED buildings can perform at the level they were designed for. As illustrated in this study and previous ones, without a monitoring and reporting mechanism in place, LEED buildings’ actual performance is questionable. To this extent, the U.S. LEED rating system can benefit from learning from other high perfor- mance building rating systems that include reporting and verification as prerequisite requirements. Zero Energy certification (by the Interna- tional Living Future Institute) and Passive House certification (by the Passive House Institute) requirements are based on a building’s actual performance. The achievement of energy goals needs to be validated after a 12-month performance period. Zero Energy certification requires building owners to report annual site energy use data, and the Passive House Institute mandates annual source energy data for certification. The Chinese Three Start Star System also requires one year of opera- tional data. Such requirements are useful tools to make sure the design, construction, and operation teams work towards the same goal. When achieving certification becomes the common goal among all stake- holders, internal conflict and split incentives can be avoided.

5. Conclusion

This study cross-referenced 2019 building energy benchmarking public data from the District of Columbia and USGBC’s LEED project inventory to compare the energy performance of DC LEED office buildings’ actual performance against that of non-LEED office buildings. The results showed that LEED office buildings do not perform better, at any of the certified levels. On the contrary, the reported LEED buildings collectively use 17% more source energy and 13% more site energy than non-LEED buildings. Among the different LEED levels, LEED Silver ap- pears to perform slightly better than other LEED levels. Meanwhile, only around 33% of qualified LEED office buildings reported their actual energy use according to the DC rule.

One of the greatest barriers to understanding the efficacy of LEED certification and other sustainable design rating systems that do not require reporting operating data has been the lack of transparent measured energy performance for commercial buildings. The USGBC has been collecting performance data for LEED buildings since 2009, but there is no indication they will make such data public [34]. Energy benchmarking programs in the District of Columbia and other metro- politan cities, such as New York, Chicago, Seattle, Austin, San Francisco, and Philadelphia, mandate making buildings’ actual performance data available to track progress in reaching the greenhouse gas emissions goal. Transparency of data has the potential to implement dramatic change by providing transparency of the energy performance of an un- precedented number of the nation’s most advanced buildings.

Declaration of competing interest

The authors confirm that there are no known conflicts of interest associated with this publication and there has been no significant financial support for this work that could have influenced its outcome.

References

[1] U.S Energy Information Administration, Use of energy explained, Accessed, http s://www.eia.gov/energyexplained/use-of-energy/commercial-buildings-in-depth. php. (Accessed 2 July 2020).

[2] A. Amiri, J. Ottelin, J. Sorvari, Are LEED-certified buildings energy-efficient in practice? Sustainability 11 (6) (2019) 1672.

[3] Wtop news, “ Mid-Atlantic remains ‘green building” leader, access, https://wtop. com/business-finance/2018/02/best-green-building-states-md-va-dc/. (Accessed 10 June 2020).

[4] District Government, Department of Energy & Environment, Final benchmarking rulemaking, Accessed, https://doee.dc.gov/node/412002. (Accessed 6 July 2020).

[5] District Government, Department of Energy & Environment, About benchmarking, Accessed, https://doee.dc.gov/node/572012. (Accessed 6 July 2020).

[6] J.H. Scofield, Efficacy of LEED-certification in reducing energy consumption and greenhouse gas emission for large New York City office buildings, Energy Build. 67 (2013) 517–524.

[7] Energy Star Portfolio Manger, Accessed, https://www.energystar.gov/buildings/ facility-owners-and-managers/existing-buildings/use-portfolio-manager. (Accessed 6 July 2020).

[8] Department of Energy & Environment, DC gov. “Building benchmarking dataset.”, Accessed, https://doee.dc.gov/publication/2017-building-benchmarking-dataset. (Accessed 6 July 2020).

[9] Open Data Dc, Building energy benchmarks, Accessed, https://opendata.dc. gov/datasets/building-energy-benchmarks. (Accessed 8 July 2020).

[10] U.S Energy Information Administration, CBECS survey data, Accessed, http s://www.eia.gov/consumption/commercial/data/2018/index.php?view=co nsumption, 2018. (Accessed 7 July 2020).

[11] J.H. Scofield, Do LEED-certified buildings save energy? Not really, Energy Build. 41 (12) (2009) 1386–1390.

[12] G.R. Newsham, S. Mancini, B.J. Birt, Do LEED-certified buildings save energy? Yes, but, Energy Build. 41 (8) (2009) 897–905.

[13] S. Energy, ENERGY STAR® Portfolio Manager™: data trends: benchmarking and energy savings, Accessed, https://www.energystar.gov/buildings/tools-and-reso urces/datatrends-benchmarking-and-energy-savings, 2012. (Accessed 24 June 2020).

[14] J.H. Scofield, Efficacy of LEED-certification in reducing energy consumption and greenhouse gas emission for large New York City office buildings, Energy Build. 67 (2013) 517–524.

[15] J.H. Scofield, J. Doane, Energy performance of LEED-certified buildings from 2015 Chicago benchmarking data, Energy Build. 174 (2018) 402–413.

[16] J.H. Scofield, O. Oberlin, A re-examination of the NBI LEED building energy consumption study, in: International Energy Program Evaluation Conference, OR, Portland, 2009, August.

[17] LEED project, accessed, https://www.usgbc.org/projects?Country=%5B%22Unite d+States%22%5D&State=%5B%22District+of+Columbia%22%5D&Rating+Sys tem=%5B%22Core+and+Shell%22%2C%22New+Construction%22%2C%22Exis ting+Buildings%22%2C%22Interiors+-+Existing+Buildings%22%5D. (Accessed 10 June 2020).

[18] U.S Energy information administration, what is RSE, Accessed, https://www.eia. gov/consumption/commercial/data/what-is-an-rse.php. (Accessed 29 June 2020).

[20] J.H. Scofield, O. Oberlin, A re-examination of the NBI LEED building energy consumption study, in: International Energy Program Evaluation Conference, OR, Portland, 2009, August.

[21] U.S. Energy Information Administration, Average size of new commercial buildings in United States continue to grow. https://www.eia.gov/todayinenergy/detail.ph p?id=21152. (Accessed 3 July 2020). Accessed.

M. Hu

Journal of Building Engineering 42 (2021) 102475

8

[22] J.H. Scofield, Efficacy of LEED-certification in reducing energy consumption and greenhouse gas emission for large New York City office buildings, Energy Build. 67 (2013) 517–524.

[23] C.M. Saldanha, S.M. O’Brien, A study of energy use in New York City and LEED- certified buildings, Proceedings of SimBuild 6 (1) (2016).

[24] J.H. Scofield, J. Doane, Energy performance of LEED-certified buildings from 2015 Chicago benchmarking data, Energy Build. 174 (2018) 402–413.

[25] P. De Wilde, The gap between predicted and measured energy performance of buildings: a framework for investigation, Autom. ConStruct. 41 (2014) 40–49.

[26] D. Calì, T. Osterhage, R. Streblow, D. Müller, Energy performance gap in refurbished German dwellings: lesson learned from a field test, Energy Build. 127 (2016) 1146–1158.

[27] U.K Government, Building performance evaluation programme: findings from non- domestic projects. https://www.gov.uk/government/publications/low-carbon-bui ldings-best-practices-and-what-to-avoid. (Accessed 5 July 2020). Accessed.

[28] https://e360.yale.edu/features/why-dont-green-buildings-live-up-to-hype-on-ene rgy-efficiency.

[29] Energy Star, What are the site-to source conversation factors?, accessed July 8, 2020, https://portfoliomanager.zendesk.com/hc/en-us/articles/216670148-What- are-the-Site-to-Source-Conversion-Factors-.

[30] O.H. Hohmeyer, S. Bohm, Trends toward 100% renewable electricity supply in Germany and Europe: a paradigm shift in energy policies. Wiley Interdisciplinary Reviews, Energy Environ. 4 (1) (2015) 74–97.

[31] G.R. Newsham, S. Mancini, B.J. Birt, Do LEED-certified buildings save energy? Yes, but, Energy Build. 41 (8) (2009) 897–905.

[32] P. MacNaughton, X. Cao, J. Buonocore, J. Cedeno-Laurent, J. Spengler, A. Bernstein, J. Allen, Energy savings, emission reductions, and health co-benefits of the green building movement, J. Expo. Sci. Environ. Epidemiol. 28 (4) (2018) 307–318.

[33] Joseph G. Allen, Piers MacNaughton, Jose Guillermo Cedeno Laurent, Skye S. Flanigan, Erika Sita Eitland, John D. Spengler, Green buildings and health, Current Environmental Health Reports 2 (3) (2015) 250–258.

[34] J.H. Scofield, Efficacy of LEED-certification in reducing energy consumption and greenhouse gas emission for large New York City office buildings, Energy Build. 67 (2013) 517–524.

M. Hu