Missing emissions control IDs leading to overestimation of NOX emissions
Closed this issue · 1 comments
When calculating controlled NOx and SO2 emissions, we rely on NOx emission rate and SO2 removal efficiency data from EIA-923 Schedule 8C ("Air Emissions Control Info"). We had been assuming that if a NOx emission rate was reported, that a corresponding NOx Control ID would also be reported, so when matching these emission rates to boilers, we relied on the NOx control ID as one of the primary keys.
However, examining our NOx results revealed they were higher than both the eGRID and EIA estimates by about 6-7%. It appears that the origin of this overestimation is that emissions control information for certain plants was not being merged because it is missing a NOx control ID, so for these plants we were calculating their emissions as the uncontrolled value.
One specific example is plant ID 10118 where we were calculating 1.4 million lb of NOx, compared to about 500 lb reported in eGRID. For this plant, the controlled NOx rates were associated with a mercury control ID, but not a NOx control ID. There are other examples where the NOx rate data is missing a NOX control ID, but is associated wth a PM control ID, SO2 control ID, or Mercury Control ID.
To fix this, we may need to load all of the different control IDs and use the other control IDs for other pollutants if the NOx control ID is missing. I am assuming that the order in which we check non-missing IDs shouldn't affect which boiler the data is associated with (ie whether we check the Hg ID first, then PM, then SO2, etc), but we should probably test this hypothesis by performing the boiler association using all of the IDs and making sure they are all the same.
This issue also affects the SO2 data, and when we add mercury or PM in the future, it would also affect those.
This also raises a question for EIA: why is there not a NOx control ID associated with the NOx emission rate data?
So after fixing this issue, it turns out that this was only affecting a small portion of the NOx emissions - our results are still significantly higher than both the eGRID and EIA data. It appears that maybe a significant portion of the difference with eGRID results from missing fuel consumption data in eGRID for certain plants. However, this does not explain why we are still much higher than the EIA estimates. This should be the next area we look into.