This is my first post in over a year. As with most of us, covid-19 has changed my priorities.

For some time now I have been studying the energy performance of LEED buildings using Municipal building energy benchmarking data.  I have also investigated the science that underpins the EPA’s building Energy Star score.  In our most recent study, these two issues intersect.

We have completed the largest study of measured energy performance of LEED-certified buildings.  The sources of our data are publicly-reported building energy data from ten cities: Boston, Chicago, Denver, Los Angeles, Minneapolis, NYC, Philadelphia, Portland, Seattle, and Washington DC.  In total we have 2016 energy data from 28,480 properties (buildings or groups of buildings) in these 10 cities.  In these data we have identified about 850 LEED-certified commercial buildings.  This has allowed us to compare the 2016 energy use of LEED buildings with those of similar buildings in the same city for the same year.

Here I want to talk about just one aspect of this study – the 2016 Energy Star scores for offices that were certified under LEED EB:OM (Existing Buildings: Operation & Maintenance).  Offices certified under a LEED EB:OM system are awarded LEED points for energy optimization based on the Energy Star score they earn during a year of operation.  This score (EAc1) can be found for many of these buildings on the Green Building Information Gateway web site.  And this score can be used to determine the Energy Star score that was reported to the USGBC used to earn this score.

Now if a building earned an Energy Star score of 85 at the time of LEED certification, you would expect it to continue earning a similar high score for subsequent years of operation.  We have compared this “Design Energy Star score” (i.e., the score that was used in awarding EAc1 points) with the Energy Star score reported in 2016 municipal benchmarking data for nearly 450 LEED certified offices in these 10 cities.  The graph below shows this comparison.  The color code indicates the LEED system used for certification.

The graph shows there is surprisingly little correlation between these two Energy Star scores.  The purple dashed line represents y = x, the expected result if the 2016 Energy Star score earned by the building is the same as that submitted for certification.  The solid black line is the actual trend line obtained by fitting a straight line to the data.

What we see is that only 19% of the variation in 2016 Energy Star scores is “explained” by the score submitted at time of certification.  81% of the variations in 2016 Scores is unrelated to this initial score.  The pink rectangle shows the range of 2016 Energy Star scores for offices that earned a score of 85 at time of certification.  There are 35 such buildings, and their 2016 Energy Star scores range from 60 to 99.  You see similar variation everywhere along the curve.  In the most outrageous cases a building’s Energy Star score in 2016 was 52 points lower than what was reported when it was LEED certified!

It is very doubtful that these changes in Energy Star score reflect real changes in the energy consumption and operation of these offices.  A much more likely explanation is that the operating parameters reported in securing these scores have changed – not reflecting real changes in computer density, or worker density, or operating hours, but just arbitrary changes due to the fact that a different person (with different motivations) reported these numbers in 2016 from the person who reported them at time of certification. The data illustrate the arbitrary nature of the Energy Star score and the ability to significantly change the score by just submitting different operating parameters.  There is no regulation of these submittals – you can change the number of computers in the building from 4000 to 8000 and there is no one who will question the number.  And this will dramatically change your Energy Star score.

In the past the EPA’s Group that works on Energy Star scores has had exclusive access to the data they gather. But with cities publishing these scores we now have the ability to check them and to see their patterns. The mean and median Energy Star score for the 4200+ offices in the benchmarking data are 74 and 80, respectively.  The EPA claims that the median Energy Star score is 50.  This is nonsense.  These 4200+ offices constitute a sizable fraction of the U.S. office space.  Like the children of Garrison Keillor’s mythical Lake Wobegon, all urban buildings are above average.