Very high emissions of Tritium to air and water for some processes

[fchichorro]

At Earthster we noticed that several processes have what appears unrealistic emissions of Tritium to water and air. The worst offender seems to be Electricity - SOLAR - South Carolina Electric & Gas Company, which has 3.03e+19 kBq of Tritium to water. But all Electricity - SOLAR, Electricity - MIXED, Electricity - COAL seem to have the same problem. The consequence is very high impacts to human health for e.g. the ReCiPe2016 impact method.

Here is a spreadsheet of processes which have Tritium emission/water for comparison.

do these levels of Tritium emissions make sense?

[m-jamieson]

Thanks for pointing this out! Honestly, I don't have a good feel for what levels of kBq make sense. In our solar model, the tritium emissions come exclusively from aluminum use in construction and the source of aluminum data is USLCI. In the older versions of USLCI (like this one) there were really high emissions/air/Tritium and emissions/water/Tritium emissions - 5.6E4 kBq and 1.1E6 kBq per kg secondary ingot. In newer versions of USLCI, I don't even see tritium listed as an emission for the entire database. We should update the data within our internal solar and wind unit models with newer USLCI data.

I'll also say you specifically call out SOLAR - South Carolina Electric & Gas, which also has really high CO2 emissions. I imagine that's because of a relatively low amount of generation for 2016 - perhaps a solar plant that was brough online in 2016 and only saw a couple months of operation. Total construction emissions are first allocated on a yearly basis based on an assumed lifetime, and then divided by the yearly generation as reported by EIA. The result is that if there's a low amount of generation in a given year, their could be really high construction emissions.

[fchichorro]

Thank you for digging this.

I tried to get a sense of what those values would mean. I found this paper reporting an inventory of 3.4e12 kBq of Tritium just after the Fukushima accident, which is way below these values, hence they look weird. We might probably just leave the Tritium out of the Inventory for these processes in our software, too.

[WesIngwersen]

This is very useful feedback, but it seems its an error in the former versions of USLCI that just needs to be propagated into these datasets.