NG end uses not correlated

[ekpresent]

Our results are showing way more homes with LP heating and nonzero NG use than RECS shows. My best guess looking at things is that this is because whether the dryer, cooking, WH, pool heater, and spa heater use NG is all dependent on heating fuel - but independent of each other, whereas in reality a home is much more likely to have multiple of those or none. Not sure what the right fix is but wanted to flag it.

[afontani]

The action would be to look at RECS 2020 and correlate these characteristics. An impact is the number of homes are applicable for upgrades. ResStock LARGEEE is showing more homes applicable compared to RECS.

[ekpresent]

I should have looped back on this back in October. This is still an issue theoretically, but the actual numbers we are seeing once I dig a little further seem to be OK. Could probably close for now, or at least assign very low priority.

This is my revised methodology for doing the comparisons, and my results.

Number of units with primary heating fuel of propane and also some NG usage:

• I used RECS 2020 microdata (downloaded 3/29/23, possibly not the latest version) and summed the NWEIGHT value of the rows that had a “2” for FUELHEAT and also a “1” for USENG. This gave me 258,468 dwelling units in this situation from 35 RECS 2020 samples.
• From ResStock, we’re showing 218,518 units with a propane furnace and some NG consumption, and 13,657 with a propane boiler and some NG consumption - so 232,175 total units (this is based on medium-scale runs, n=30,000)
• I feel these numbers match up reasonably well between ResStock and RECS.

Number of units with primary heating fuel of fuel oil and also some NG usage:

• Same methodology as above but “3” for FUELHEAT and also “1” for USENG. This gave me 944,636 dwelling units in this situation from 139 RECS 2020 samples.
• From ResStock, we’re showing 651,003 dwelling units with a fuel oil furnace and some NG consumption, and 418,827 with a fuel oil boiler and some NG consumption - so 1,069,830 total units (again, from an n=30k run).
• These numbers seem within range to me.