Interpretation of steam data from CEMS

[grgmiller]

We had been assuming that the reported hourly steam load reported in CEMS represented steam that was used for heating (as part of a CHP plant) and not for electricity generation. However, a more careful read of the EPA power sector data user guide suggests that this steam could just represent steam generated as part of a combined cycle turbine. Instead of dropping this data when adjusting for electricity, we might want to keep it in. Instead, we may want to cross-reference with information reported in EIA-923 about whether the plant is CHP, and whether some portion of the fuel input ultimately goes toward heat rather than power.

[grgmiller]

On further inspection, it appears that the steam load numbers reported by CEMS might not be directly related to useful thermal output from the plant, and should probably be ignored.

Thus, we need to adapt our method for how to calculate the fuel consumed for electricity, and how to calculate the adjusted emissions.

Calculating Fuel Consumed for Electricity

Previously, our approach had calculated this based on the heat-content (mmbtu) weighted ratio between gross_generation_mwh and steam_load_1000lb in each hour. However, this approach assumed that steam load represented thermal output, and that none of it was used in combined-cycle power generation.

Our new approach uses the data in EIA-923 and appears to be more consistent with the methodology used by eGRID. Our specific steps are to:
1. Calculate the ratio between fuel_consumed_for_electricity_mmbtu and fuel_consumed_mmbtu in EIA-923 for each subplant-month
2. Calculate the same ratio at the plant-month level instead of the subplant-month level
3. Merge these fuel ratios into the CEMS data
4. Use this ratio to calculate a fuel_consumed_for_electricity_mmbtu from the fuel_consumed_mmbtu data reported in CEMS. If available, we use the subplant-month specific ratio. Otherwise, we use the plant-month specific ratio. Otherwise, we assume a ratio of 1 (essentially that there is no allocation to heat output).

Calculating the electric allocation factor and adjusted emissions

This approach remains mostly the same as before. However, previously, we had used the CEMS gross generation in the electric allocation factor formula, instead of using net generation, as is used when doing this calculation for the EIA-923 data. To ensure greater consistency, we now use the net generation number instead of gross generation. This means that we had to move these adjustment calculations out of data_cleaning.clean_cems() and into a separate function that is executed after we calculate net generation for the CEMS data.

Calculating adjusted fuel consumption for electricity

So it turns out that eGRID not only adjusts emissions using the electric allocation factor, but also re-calculates fuel_consumed_for_electricity as well by multiplying the electric allocation factor by the total fuel consumption.