search

EnergyInnovation / eps-us

Energy Policy Simulator - United States
GNU General Public License v3.0
22 stars 7 forks source link

Fix incorrect value for NG feedstocks in refining industry in BPoIFUfE #191

Closed jrissman closed 3 years ago

jrissman commented 3 years ago

When the Industry Category subscript was expanded to include more industry categories (for EPS 3.2.0, in commit 742ed701e4393a155d58e7b32501b2735dfa1a44), the variable indst/BPoIFUfE contained data from the AEO reference scenario. In Table 24, that includes a breakout of natural gas use by the refining industry that specifies the shares of natural gas used for heat and power vs. feedstocks.

Subsequently, when the model was updated to use a different EIA AEO scenario as our BAU case, this file was changed to a version in which the EIA has left those breakout cells blank. This created an erroneous value of zero for the fraction of natural gas used by the refining industry for heat and power. (The real value is around 85%.) This happened prior to the public 3.2.0 release, so we have never actually had an EPS release with the expanded Industry Categories that did not contain this error.

If any U.S. state or international models with version number 3.2.0 or later have inherited BPoIFUfE from the U.S. national model, they also contain this error and should be updated.

I plan to fix this one myself because I need the correct data today.

jrissman commented 3 years ago

I've completed this in b0ae6d1. Are there any state models that inherited BPoIFUfE from the U.S. national model? If they are for a state that contains significant refining (such as Louisiana or Texas), it might be important to update those U.S. state models.

jrissman commented 3 years ago

The U.S. state models were actually using a value of 100% for energy, based on MECS. RMI inquired with EIA and determined that 85% is correct, since MECS omitted this feedstock use, as it appeared on some other survey instead. RMI will be updating the state models to use the 85% figure.

We should be cautious about assuming MECS has a complete breakout of all industrial energy use data in the future. I've become less happy with MECS as a data source as the gaps in it start to become clearer. AEO remains the best U.S. source for present-day industrial energy use, with MECS most useful when a greater disaggregation in end uses is needed (like boiler fuel vs. process heat needs within a particular industry).

Closing this issue.