Work demonstrating variability in LCIA results for ecoinvent systems outside of parametric uncertainty. The work uses lca-tools to compute dynamic range of market mixes / review market mixes and brightway2 is used for Monte Carlo simulations of ecoinvent. MCS runs are saved by default to disk as json files by UUID, so that a statistical record will continue to grow until the file is deleted.
To run catalog-based first:
- '0. Setup Catalog.ipynb'
- 'DEMO-0'
To run brightway 2:
- 'DEMO-1' and just have ecoinvent named the same way I do
Then:
- 'DEMO-2' does demonstration figures comparing the discrete and continuous probability distributions
- 'DEMO-3' measures the extent / utilization of markets in ecoinvent (conseq left out for some reason :( )
- 'Editorial image' does the figure from the editorial.
- 'Freight transport study' ongoing for paper- but it seems optimistic for a GLO model (bc pedigree 12241 or whatever)
Other stuff I should just delete.
Three main things:
- lca_variability - lca-tools catalog-based study of market variability, based on input flows matching reference flows
- bw2_mca - Brightway 2 hack to run multiple LCIA methods per MCS run
- bw2_mkt_weight - subclass to implement choice among suppliers based on market weight
- And then the charts - stem and hist mainly.
four things.