Add variability in ground thermal conductivity

[afontani]

Pull Request Description

Companion PR: resstock-estimation PR #393

Allow ground conductivity to vary by climate zone. The GeoVision task force report provided data by climate zone. The same data is used to fill out distributions by climate zone.

Data source

• Constructed using data from the SMU Geothermal Laboratory.
• The data is from the Thermal Conductivity Observation in Content Model Format dataset.
• The data is available at http://geothermal.smu.edu/static/DownloadFilesButtonPage.htm.

Assumptions:

• The data obtained is from surveyed oil and gas well data.
• The latitude and longitudes were assigned to counties and the data was joined to the ResStock spatial lookup tables. In this process, 1482 of 59332 samples did not have a FIPS match or did not have data and were dropped.
• Due to limited data in climate zone 1A, data was pulled from samples in 1A plus Florida 2A.
• Samples less than 0.5 Btu/hr-ft-F are assigned a value of 0.5 Btu/hr-ft-F. Samples greater than 2.6 Btu/hr-ft-F are assigned a value of 2.6 Btu/hr-ft-F.

Raw Data Distributions by IECC Climate and Moisture Zone

Checklist

Not all may apply:

• [x] Tests (and test files) have been updated
• [x] Documentation has been updated
  • [ ] If related to resstock-estimation, checklist includes data dictionary, source report, options saturation, options_lookup.
• [x] Changelog has been updated
• [x] openstudio tasks.rb update_measures has been run
• [x] No unexpected regression test changes on CI (checked comparison artifacts)

[joseph-robertson]

@afontani I suppose the addition of "Ground Thermal Conductivity.tsv" reshuffles samples? So, using comparisons of results from CI, it's hard to get an idea of impact here? I wonder if it makes sense to do a base/feature comparison where a fixed buildstock.csv is used (i.e., introduce the new tsv to the base branch, but the lookup assigns the original defaulted value of 1.0)?

[afontani]

@afontani I suppose the addition of "Ground Thermal Conductivity.tsv" reshuffles samples? So, using comparisons of results from CI, it's hard to get an idea of impact here? I wonder if it makes sense to do a base/feature comparison where a fixed buildstock.csv is used (i.e., introduce the new tsv to the base branch, but the lookup assigns the original defaulted value of 1.0)?

@joseph-robertson : Yeah, any characteristic changes shuffle samples and the artifact comparisons are less useful. Their aggregate impact should give a sense of what the changes are. Let me create a couple plots.

[afontani]

@joseph-robertson : I took a look at bar charts using the datapoint base and feature results. I did not see significant changes outside the uncertainty expected with this sample size. Small increases in delivered heating and a small reduction in delivered cooling. This is expected as the ground conductivity is increasing.

Examples: