Closed CarrerF closed 1 year ago
Good catch. I think it is worthwhile to go through these issues when they produce undesirable or unrealistic outcomes. I would suggest that we look at ways to import data from new scenarios, e.g. the SHAPE scenario explorer. We are free to choose any of the models and do not need to stick to Message. However, using newer scenarios means that more people have worked on issues such as these from the IAM side.
I agree! Although there have been quite a few updates and re-downloads of the MESSAGE-ix CD-LINKS project results. That parameter is among those with the highest number of revisions (more than 10).
Just to clarify: These emissions are supply chain emissions only from the generation/production of the individual fuels. The direct emissions from burning the coal, natural gas, etc, in one of our system's processes are calculated separately with 6_PR_DirectEmissions_V1.2
My suggestion is that you use your SHAPE project insights, also with the LCA method link, to explore improvement options for the GHG factors per unit of energy.
Thanks!
Before opening this issue, I have been exploring the possibilities of using SHAPE results.
From IMAGE:
I would be careful using IAM results as input for RECC.
The current implementation of energy supply (4_PE_ProcessExtensions_EnergyCarriers_V1.0) has no time and R-scenario aspect for the individual technology.
That means, currently, the electricity mix changes over R and t but not the emissions intensity of the individual technologies.
@CarrerF , is there an update on this pending? Shall we keep that simplification or shall we, for RECC v2.5, switch back to the MESSAGE energy supply results and invest a bit of time in understanding and fixing them?
current implementation:
# replicate 2015 values. The current dataset contains only the 2015 initial value.
# Electricity is added separately in the next step
ParameterDict['4_PE_ProcessExtensions_EnergyCarriers_MJ_o'].Values = np.einsum('nxo,n,tR->nxotR', ParameterDict['4_PE_ProcessExtensions_EnergyCarriers'].Values[:,:,:,0],ParameterDict['4_EI_SpecificEnergy_EnergyCarriers'].Values,np.ones((Nt,NR)))
ParameterDict['4_PE_ProcessExtensions_EnergyCarriers_MJ_Materials'].Values = np.einsum('nxo,n,tRm->mnxotR',ParameterDict['4_PE_ProcessExtensions_EnergyCarriers'].Values[:,:,:,0],ParameterDict['4_EI_SpecificEnergy_EnergyCarriers'].Values,np.ones((Nt,NR,Nm)))
Also, there is currently no impact for hygrogen. We can at least link hygrogen to electricity supply.
Switching back to IAM results is a possibility, if enough time can be invested (back to https://github.com/IndEcol/RECC-ODYM/issues/39#event-9184452530)
Also, IAM scenarios consistent with these GHG trajectories for environmental impacts other than GWP are missing.
Allright, I'll think of a quick way to estimate hydrogen production impacts (likely going via the electricity demand), and then, we are all set for new results, and can iterate later on other improvements. But for now, all looks good, incuding the material footprint and process emissions calculations, really consistent, high quality and well documented work Fabio!
PS: I think the best quick solution is to introduce a parameter "electricity demand of H2 production in MJ/kg" and loop back the H2 production to connect to (average) electricity generation.
The quality of the dataset should be discussed: Quite often, the trajectories have
increasing values even for SSP1 scenarios
![incr (2)](https://user-images.githubusercontent.com/98351491/177200088-47e31e74-8a36-4d50-ade7-fc100f5dbad8.png)
RCP scenario have higher values than baseline (and late century explosion)
![high RCP (3)](https://user-images.githubusercontent.com/98351491/177200181-32955b29-fce9-49d3-bbbf-59108d3b441a.png)
unrealistic trajectories![traj](https://user-images.githubusercontent.com/98351491/177200242-f010bc28-b29e-4248-b6a2-536d3116ac32.png)
I had a look a the original MESSAGE data and the corresponding parsing script. The GHG intensity is computed by dividing the total emissions by the final energy. The original trajectories are jumpy already, and scattered in time, which doesn't help. The energy carrier - final energy mapping is then done manually in the parsing script.
I'm especially concerned about the late-century explosions of trajectories. I've questioning myself about the trajectories for some individual countries at the end of the time horizon [example: REB for France]![Figure 14_GWP_TimeSeries_Materials_Stacked_France, LED, RCP2 6 energy mix png_France](https://user-images.githubusercontent.com/98351491/177202099-ac56659e-bace-4f64-8810-6f3ffce6b3b6.png)