methodology

The overall goal of this analysis was to identify and segment a limited number of building typologies that would collectively encompass the largest portion of the office building stock in New York City, by floor area and greenhouse gas (GHG) emissions. The publicly available NYC Local Law 84 (LL84) energy benchmarking dataset was leveraged to quantify the gross square footage (GSF), count, and total GHG emissions of office buildings in New York City between 25,000 sf and 1 million sf. The publicly available Local Law 87 (LL87) energy audit dataset was leveraged to identify the most common building systems in those office buildings, and then segment out four typologies, based on major heating and cooling systems, for further analysis. The analysis required examining data for every individual building from both datasets, and then cleansing the LL84 data and LL87 data to filter for buildings that are included in both datasets, to verify that whole building energy use and GHG emissions data energy use data reported in LL87 audit reports is indeed for the whole building.

Local Law 84 Data Cleansing

To cleanse the LL84 dataset, the team started with data from 2018, which included 25,245 submissions. They then completed the following steps:

1. Removed buildings with no office space typologies (Office, Financial Office, Medical Office, and Veterinary Office) in the following columns: Primary Property Type—Self Selected; Primary Property Type—Portfolio Manager Calculated; Largest Property Use Type; 2nd Largest Property Use Type; and 3rd Largest Property Use Type.
2. Removed properties with less than 50k sf and properties with more than 1M sf.
3. Removed properties with EUI >1000 or <5 [kwh/sf/yr].
4. Removed properties with no electricity use.
5. Removed properties with estimated energy values.
6. Removed properties with less than 12 full calendar months of data.
7. Removed properties with district chilled water > 0.
8. Removed duplicate property IDs, leaving the most recent submission.
9. Removed properties with duplicate Building-Block-Lot (BBL) and Building Identification Number (BIN), leaving the most recent submission.

Following this data cleansing process, the team was left with a total of 2,257 submissions.

Local Law 87 Data Cleansing

To cleanse the LL87 dataset, the team started with data from 2013 to 2018, which included 7,083 submissions. They then completed the following steps:

1. Removed properties with less than 50% office square footage. Acceptable office space typologies were: Financial Office, Office, Medical Office, and Veterinary Office. Following this step, 773 submissions remained.
2. Removed buildings below 50,000 sf, leaving 769 remaining.
3. Removed buildings over 1M sf, leaving 712 remaining.
4. Removed buildings without a BBL match between LL84 and LL87 data sets, leaving 583 remaining.
5. Removed duplicate BBLs and kept the most recent submission, leaving 566 remaining.
6. Removed buildings where the space type reported in LL84 and LL87 data did not match, leaving 546 remaining.
7. Remove buildings with site EUI below 5 or above 1000 [kwh/sf/yr], leaving 479 remaining.
8. Remove buildings with a total site EUI discrepancy of 50% or more between LL84 and LL87 data (to account for false reporting), leaving 354 remaining.
9. Remove buildings with a discrepancy of 20% or more in GFA between LL84 and LL87 data, leaving 320 remaining.

After completing this data cleansing process, the team segmented the remaining 320 office properties into four building typologies, based on primary heating and cooling systems, in order to then calculate end use energy breakdowns, as well as Energy Conservation Measure (ECM) savings. This left the team with a total of 263 buildings across the four chosen typologies.

ECM Data Cleansing

To accurately calculate potential savings that can be achieved by implementing LL87 energy audit recommended ECMs, the team began by examining LL87 ECM submissions from 2013-2018, which included 47,924 ECM recommendations. The team then took the following steps to cleanse the data:

1. Filtered to select office property type.
2. Removed duplicate submissions.
3. Removed ECMs with data type issues (e.g., text listed in an important numerical column).
4. Removed ECMs with missing energy savings.
5. Removed ECMs with missing cost savings.
6. Removed ECMs where the total energy savings are higher than the total building energy consumption.
7. Removed ECMs where the energy savings = 0.
8. Removed ECMs where the cost savings = 0.
9. Removed ECMs where the energy savings are negative.
10. Removed ECMs where the cost savings are negative.
11. Removed ECMs where the calculated total savings differ from the self-reported total savings by more than 30% (this is to catch ECMs where the individual fuel savings were reported in units other than kBtu as units were not provided in the LL87 data).
12. Remove any ECMs from properties that did not pass the LL87 data cleansing process.

Following this process, the team was left with a dataset of 2,507 total ECMs, from which they could calculate savings
by end use system.

End Use Breakdown Methodology (Energy & GHG)

To calculate the end-use breakdown (EUB) by building typology, we utilized the set of 263 buildings that passed the LL87 cleansing above. To calculate the GHG associated with each EUB category, we calculated the GHG emissions per EUB category for each property. To do that, we took used the follow steps and assumptions:

1. Multiplied each EUB category’s consumption by the relevant GHG emissions factor
2. GHG emissions factor was based on the assumed primary fuel type for that building and EUB category.
3. The space heating and domestic hot water factors were based on the primary heating fuel type defined for the building
4. The space cooling factor was based on the primary cooling fuel type defined for the building, and the other categories used the electricity emissions factor.
5. The emissions factors used were the ones outlined in Local Law 97.

Savings Potential by End Use Category for Each Typology (Energy & GHG)

To calculate the ECM savings potential by end use category, we started with the ECM dataset as defined in the ECM Cleansing Methodology above. This left us with 2,507 ECMs from which to calculate potential savings across 263 properties. We then used the following steps and assumptions:

1. Filter out ECMs where the energy savings are greater than the reported end-use-breakdown category, leaving us with 2,225 remaining ECMs.
2. Calculate the GHG savings for each ECM by multiplying the individual fuel type savings by the GHG emissions factor for that fuel type as defined by Local Law 97.
3. Sum the total savings for each of the four office typologies by ECM category (this is different than the EUB categories, LL87 does not have you specify which EUB category your savings came from).
4. Use the total savings by ECM category to calculate the total savings by EUB category using the following methodology:
   a. For each ECM category that applies to only one EUB category, add the summed value for that ECM category to the EUB category sum
   b. For each ECM category that applies to more than one EUB category, split the ECM savings into the respective EUB category, weighting the savings by the EUB consumption. For example, if an ECM category applies to space heating and  space cooling, and 2x the energy is used for heating vs cooling, assign 2/3 of the savings to space heating and 1/3 to space cooling.
5. Copy the total consumption (energy & GHG) by EUB category for each typology as calculated in the previous section.
6. Calculate the post-savings consumption as total consumption minus estimated savings.
7. Calculate the savings as a percentage of the total consumption as the estimated savings for that category divided by the total consumption for that typology.
8. Finally, we calculate the savings as a percentage of the EUB category as the estimated savings for that category divided by the total consumption for that category.

Citywide Extrapolation of Typologies and GHG Savings

Finally, to estimate the citywide savings potential for the identified energy reduction measures for the four typologies we extrapolated our findings from the cleansed LL87 data and applied them to the LL84 data set of building between 25,000sf and 1 million sf.

The proportional count, total area, GHG emissions and GHG savings for each typology within the cleansed LL87 data was applied to the universe of office buildings between 25,000sf and 1 million square feet from the 2018 LL84 data set.

Building Typology Assignment

Buildings were assigned to one of four typologies based on their primary heating system, primary heating fuel, primary cooling system,
and whether they have centralized or decentralized cooling.

Heating Fuel and System Assignment

Heating fuel typologies were simplified from the list provided in the LL87 submissions to the list below, using the following mapping:

Equipment Assignment

The centralized vs decentralized cooling was determined from the primary cooling system using the following mapping: