How does ResStock calculate conditional probability distributions in housing_characteristics tsv files?

asked 2020-06-25 15:38:51 -0600

sashadf1 gravatar image

updated 2020-06-25 17:41:51 -0600

For reference I am using ResStock v2.2.4 and OpenStudio/PAT v2.9.0, with EnergyPlus v9.2.

I am trying to model a particular community in Georgia. I have 3 different AMY EPW files that I have purchased from Whitebox Technologies: a 2008 EPW file for Blairsville GA, a 2008 EPW file for Andrews NC, and a 2008 EPW file for Franklin NC.

In order for PAT to recognize these EPW files, I have had to rename the EPW files as follows before zipping them and uploading them to dropbox to obtain a direct download link for input into PAT's server script section, depending on what Location I select out in Location.tsv within the housing_characteristics directory. The options I have determined to be candidates in Location.tsv are Asheville AP (NC), Athens AP (GA), and Chattanooga AP (TN), based on their proximity to the desired location I am modelling.

Location TSV configurations for Asheville AP and Chattanooga AP are presented below. TSV configurations are mutually exclusive, as in I would only run one PAT simulation per configuration at a time. I suspect an error would be thrown if there were multiple "1s" in Location TSV file.

Asheville Location TSV config

Chattanooga Location TSV config

In order for PAT to recongize the EPW files, their names were as follows depending on what Location.TSV config I was running with.

Asheville Location TSV configuration --> 0996216_US_NC_Buncombe_37021_0-20000-0-72315_35.45_-82.54_NSRDB_2.0.1_AMY EPW file name.

Chattanooga Location TSV configruation --> 0939546_US_TN_Hamilton_47065_0-20000-0-72324_35.05_-85.22_NSRDB_2.0.1_AMY EPW file name.

Athens Location TSV configuration --> 0979938_US_GA_Clarke_13059_0-20000-0-72311_33.97_-83.34_NSRDB_2.0.1_AMY EPW file name.

Thats the background. Here's the real issue.

I noticed that if I ran two simulations, all else held equal INCLUDING APPLY UPGRADE MEASURE AND THE EPW FILE USED, I would get different results depending on what configuration I was using in Location TSV. I did some digging into specific files in housing_characteristics, and I noticed that Geometry Foundation Type.tsv and Heating Fuel.tsv had Dependency=Location, meaning that their conditional probability distributions are conditional on whatever value/configuration is in Location.tsv. When I changed these conditional probability distributions to possess the same values, I would get the same results.csv output.

For example, see:

Geometry Foundation Type TSV change

Heating Fuel TSV before edit

Heating Fuel TSV after edit

How does ResStock calculate the probability distributions seen above for Heating Fuel and Foundation Type? I would like to get a probability distribution that is representative of my area. I was able to insert my own custom distribution for vintage.tsv: see below. This distribution was calculated using ArcGIS and survey data. I would like to be able to create something similar for Geometry Foundation Type and Heating Fuel tsv sections.

Vintage TSV custom distribution

edit retag flag offensive close merge delete