Currently, HEAT uses the approach described as Option 3 in this Google doc to set the default Design Temperature. To summarize, through geocoding of the home's address, the HEAT determines the county where the home is located, and then looks up the design temperature in a table that includes all U.S. counties. The table was obtained from a U.S. DOE publication. Unfortunately, the methodology in that publication results in design temperatures that are substantially lower than those recommended by ASHRAE, which is considered the golden source.
An alternate method that was used by Thermentor would be to download 8-to-10 years of hourly temperature data from open-meteo.com and compute the design temperature. Thermentor told me that they used only 8 years, but I believe the ASHRAE standard uses 10 years. I would like someone to first test this approach for the 6 weather stations in the original spreadsheet. This could be done using just Python, even separately from our code base. The table this test would complete is the following, where 8-year and 10-year calculated are computed from temperature data downloaded from open-meteo for 8 and 10-year periods ending 12/31/2025:
** Weather ------------Heating Design Temperature (degF) ------------ Query & calc time (s)
**Station code ASHRAE 99% temp 8-yr calculated 10-yr calculated 8-yr 8-yr 10-yr
KBED 8.4
KBOS 13.4
KBVY 9.5
KFIT 7.7
KORH 7.1
KOWD 9.0
For each weather station, use Google Maps to find two home addresses -- one should be very close to the weather station and a second should be 1/3 to 1/2 of the distance to one of the other weather stations in this list. Create a test that completes two copies of this table (one for homes very close to the weather station and a second for homes part way between two weather station). The test should also measure the time for data point. The test could be implemented just in Python and completely independent from the rest of the code.
I used CoPilot AI to get more details on how ASHRAE computes heating design temperature. CoPilot's response seems reasonable to me, so please follow that in doing the calculation.
Please ask Steve B to review the two tables. If Steve determines that this new approach produces reasonable results for the 6 weather stations listed above, the next step is to proceed with implementing this approach in HEAT.
Currently, HEAT uses the approach described as Option 3 in this Google doc to set the default Design Temperature. To summarize, through geocoding of the home's address, the HEAT determines the county where the home is located, and then looks up the design temperature in a table that includes all U.S. counties. The table was obtained from a U.S. DOE publication. Unfortunately, the methodology in that publication results in design temperatures that are substantially lower than those recommended by ASHRAE, which is considered the golden source.
An alternate method that was used by Thermentor would be to download 8-to-10 years of hourly temperature data from open-meteo.com and compute the design temperature. Thermentor told me that they used only 8 years, but I believe the ASHRAE standard uses 10 years. I would like someone to first test this approach for the 6 weather stations in the original spreadsheet. This could be done using just Python, even separately from our code base. The table this test would complete is the following, where 8-year and 10-year calculated are computed from temperature data downloaded from open-meteo for 8 and 10-year periods ending 12/31/2025:
** Weather ------------Heating Design Temperature (degF) ------------ Query & calc time (s)
**Station code ASHRAE 99% temp 8-yr calculated 10-yr calculated 8-yr 8-yr 10-yr
KBED 8.4
KBOS 13.4
KBVY 9.5
KFIT 7.7
KORH 7.1
KOWD 9.0
For each weather station, use Google Maps to find two home addresses -- one should be very close to the weather station and a second should be 1/3 to 1/2 of the distance to one of the other weather stations in this list. Create a test that completes two copies of this table (one for homes very close to the weather station and a second for homes part way between two weather station). The test should also measure the time for data point. The test could be implemented just in Python and completely independent from the rest of the code.
I used CoPilot AI to get more details on how ASHRAE computes heating design temperature. CoPilot's response seems reasonable to me, so please follow that in doing the calculation.
Please ask Steve B to review the two tables. If Steve determines that this new approach produces reasonable results for the 6 weather stations listed above, the next step is to proceed with implementing this approach in HEAT.