Approach for ODS Substitutes from refrigerants

[bl-young]

This relates to emissions from Table 4-124

[bl-young]

CEDA utilizes total "Refrigeration/Air Conditioning" emissions and attributes across all sectors based on Use table data for purchase of Air conditioners (333415).

Flowsa relies on the individual sub categories:

Mappings are specific for each sub category - this results in a fairly significant reallocation between commercial / industrial / residential (presumably the Use table split between these large sectors does not match the emissions reported in GHGI)

[bl-young]

The adjustment to attempt to differentiate by gas is also different, where CEDA adjusts the CO2e values based on changes between AR5 and AR6, and makes some assumptions regarding which pollutants stem from which activity:
https://github.com/cornerstone-data/bedrock/blob/5db90cdaa7666856c9cae4cd67ec258ed6080121/bedrock/extract/allocation/epa.py#L1014-L1050

While in Flowsa, the breakout of HFC type is assumed the same for all activities.

[WesIngwersen]

EPA used the Vintaging Model (VM) for these estimates which is a highly detailed bottom-up model of using of ozone depleting substances (ODS) substitutes (which is generally HFCs) across equipment types with estimates of chemical loads and loses across different equipment phases/time. Details are in the Appendix to the US GHGI. Through this model the data are available by specific GHG by equipment type but the total loads and emissions for specific equipment are not published in the Appendices as (I assume this is the reason) the model does rely on some confidential business information (CBI).
Presentation here: https://climate-transparency-platform.org/sites/default/files/2024-06/D5_ODS_Dave%20Godwin.pdf

It would be difficult to use the data from the model directly even if we had access because NAICS based assignment is not always clear on this equipment, but individual gases would get resolved by equipment.

[damienlieber-dnexus]

Here are my observations after deep dives into the contribution of refrigerants in the direct emissions of USEEIO sectors.

Experts that are part of the superpollutant action initiative noted a few surprisingly high contributions, like management consulting (90%, see https://decarbnexus.github.io/useeio_ghg_sources_disaggregation/). However, major management consulting firms typically don't report refrigerants in their scope 1 because the emissions are "immaterial" (e.g., Deloitte p27). I've also heard from experts that emissions from leaks, servicing, and disposal should be small for large commercial and industrial users because they and their contractors know the regulations and processes to avoid them. This is all anectodal and repeated from others, but I wanted to share if useful and I'm happy to introduce you to the carbon containment lab's experts to further explore these potential discrepancies.

On the other hand, I've noted that other sectors have surprisingly small contributions from refrigerants, like "Seafood product preparation and packaging" (0.1%) or "Cheese manufacturing" (1%). I'd expect a lot of refrigeration in those operations, but most of their emissions come from Natural Gas Combustion for Manufacturing Purposes.

These potential discrepancies could be due to under/overallocation of refrigerant use and/or conversely, over/underallocation of all other GHG sources. Have you diagnosed similar problems previously and if so, could we replicate that approach? Otherwise here are a few ideas I thought of so far:

• We could compare the EPA's VM model / GHGI results for refrigerant use emissions by sector with the use table data of refrigeration and AC equipment by sector to investigate how correlated they are
• We could benchmark a few sectors based on corporate disclosures and investigate how correlated they are
• We could explore alternative attribution data sources with experts in refrigerant use from the carbon containment lab or other institutions