Add emissions from LNG terminals

[mkmahajan]

We don't currently break this out, since we don't typically track impacts of exported fuels (though methane leakage from current LNG facilities are included in the EPA Inventory and therefore included in our methane estimates). But it would be helpful to separately break this out and be able to correctly track the methane impacts of LNG terminal expansions.

[jrissman]

We do track CO2 impacts of exported fuels. We have graphs of it in the web tool - see the "Imports, Exports, Consumption, and Production" category, graphs "Energy: Embedded CO2 in Exports" and "Energy: Change in Embedded CO2 in Exports." They are already broken out by fuel type, so you can see natural gas there.

It doesn't distinguish between natural gas exported by tanker ship versus by pipeline because the natural gas is chemically identical, and it makes no difference what mode of conveyance was used to export the natural gas. I don't think we should support that distinction in the model code. A user who only cares about NG that is exported by ship can simply multiply the CO2 embedded in exported NG (as reported by the model) by the percent of NG they believe is exported by ship. (And if they are looking at LNG terminal expansions, they don't even need to do this - all they need to know is how much more LNG they expect the terminal to export.)

If you are looking at emissions from the terminals themselves, not from the exported fuels, that is not something we break out. Aren't the emissions from terminal operations too small to be worth breaking out? Around 90% of methane leakage occurs at oil and gas fields, rather than at later points in the natural gas transmission and distribution system or end use equipment, according to EPA's global non-CO2 inventory report. Does the EPA inventory break out methane leaked from LNG terminals so we can see what percentage it is of the total methane leakage?

[jrissman]

Note that the EPA global non-CO2 report could be wrong. For instance, this study found that methane emissions from fertilizer plants are 100 times higher than reported to the EPA, so EPA data could be under-counting methane from other types of operations as well.

[robbieorvis]

The issue is not with the embedded CO2; it is with the terminals themselves.

Each one of the terminals is like a 500 MW coal plant; the emissions are not trivial. This is because in addition to onsite leakage at the terminal itself, there is a significant amount of combustion of the gas to liquefy it. There are also significant local air quality issues from NOx.

LNG is an important topic in the current discourse, and I think it would make sense to add a little bit of structure to capture this better. It would probably just require either adding a variable specifying the share of natural gas exports that are via pipeline or terminal (perhaps this is in a timeseries file or setting caps for certain export types (e.g. pipeline) and having the model allocate the rest. Then we just need a variable that specifies the fraction of exported gas that is combusted (and adjusts exports accordingly) and we can use the industry sector PEI values to estimate emissions.

It need not necessarily make it into 4.1, but since it is a hot topic now and the foreseeable future, it would be a nice thing to have.

-R

[jrissman]

Okay, makes sense to me!

Incidentally we could consider not following the EPA inventory if we think they are vastly undercounting methane leakage, and if we think deviating from the inventory would not cause problems with policymaker acceptance and/or comparability with other models or datasets.