Fix fast export for `brightway2`

What's Changed

• Codex/fix bw2 process type export by @romainsacchi in #291

Full Changelog: v.2.4.1...v.2.4.2

Preserve metadata in fast Brightway exports

Fix an issue detected by @tomterlouw where exchange attributes and metadata were dropped when fast-writing databases to Brightway.

RSS optimized

Highlights

Faster and lighter workflows

• Much faster and lighter write_db_to_brightway() on Brightway 2.5
• Lower RAM usage during cold NewDatabase() initialization
• Leaner post-update() scenario caching and reload
• Faster Brightway superstructure export

Better Brightway export behavior

• Fast exports now behave much more like regular Brightway databases
• Exported databases remain searchable and browsable
• Activity Browser compatibility was fixed by ensuring process types are set
• Export UX is clearer, with progress bars, overwrite notices, and an explicit “database
  written” message

Updated data and assumptions

• Refreshed solar PV efficiency assumptions
• Added new PV anchor years and a perovskite-Si tandem trajectory
• Expanded GCAM variable mappings across sectors
• Updated GCAM regional mapping assets, including Ukraine coverage

Workflow and data fixes

• Better IAM normalization when an exact 2020 datapoint is missing
• End-of-pipe emissions updates moved later in the workflow to avoid ordering issues
• Improved electricity validation diagnostics
• Fixed cold-cache reload behavior
• Fixed migrated inventory gap filling, where replacement-based exchanges could be
  duplicated incorrectly after migration splits

In one sentence

2.4.0 moves premise from a mostly correct but heavy Brightway export workflow to a much
leaner one, while also updating PV and GCAM assumptions and fixing several export, cache,
and inventory edge cases.

Observed local benchmark

On a local full three-scenario cold workflow benchmark, compared to 2.3.9:

• runtime dropped from about 2564 s to about 717 s
• peak RSS dropped from about 4.38 GiB to about 2.03 GiB

Remove soft dependency to Brightway

Allows creating Simapro and OpenLCA-compatible database without needing a Brightway project or Biosphere database.

Update Zenodo link to access update IMAGE scenarios

v.2.3.8

update zenodo link to new IMAGE scenarios

MESSAGE scenarios

Added

• Added MESSAGE scenarios.
• Region-to-feedstock mapping files for biofuels:
  • premise/iam_variables_mapping/iam_region_to_biodiesel_feedstock.yaml
  • premise/iam_variables_mapping/iam_region_to_bioethanol_feedstock.yaml
• Metals post-allocation correction files for newer ecoinvent versions:
  • premise/data/metals/post-allocation_correction/corrections_311.yaml
  • premise/data/metals/post-allocation_correction/corrections_312.yaml

Changed

• Expanded fuel regionalization, including higher-level fuel activities.
• Expanded biomass supply-chain regionalization.
• Updated IAM variable mappings across sectors (final energy, fuels, heat, transport, electricity, steel, CDR).
• Updated GCAM topology and related mapping assets.
• Updated Brightway export/import integration (premise/brightway2.py, premise/brightway25.py).
• Updated inventory assets:
  • premise/data/additional_inventories/lci-biofuels.xlsx
  • premise/data/additional_inventories/lci-final-energy.xlsx
  • premise/data/additional_inventories/lci-pass_cars.xlsx
• Updated SimaPro category and import classification mappings.

Fixed

• Multiple MESSAGE mapping fixes (fuels, final energy, industrial heat).
• Duplicate regionalization / duplicate dataset creation issue.
• Improved missing-file error messages.
• Unit normalization for IAM inputs provided in PJ/yr (conversion to EJ/yr).
• Updated NOx and PM2.5 emission factors for diesel passenger cars.
• Reintroduced SSP2-Base as an accepted scenario.

Data cleanup

• Removed obsolete fuel config files:
  • premise/data/fuels/fuel_efficiency_parameters.yml
  • premise/data/fuels/fuel_markets.yml
  • premise/data/fuels/hydrogen_supply_losses.yml
  • premise/data/fuels/supply_chain_scenarios.yml
• Renamed metals correction file:
  • premise/data/metals/post-allocation_correction/corrections.yaml
  • to premise/data/metals/post-allocation_correction/corrections_310.yaml

Documentation

• Updated README.md, multiple docs pages, and example notebooks.

Fix non-regionalized fuel supply chains

Since probably 2.3.0, fuel supply chains ceased being regionalised.

Fixes a few bugs

• Add header in Zenodo requests to prevent being denied
• Add guard to checking parameters in dataset during validation to prevent missing key error
• Pin down pandas<3.0.0
• Redeclare extract_brightway2_databases as wurst.extract_brightway2_databases

New migration system

• Ecoinvent 3.12 compatibility
• Improved migration system, relying on bw_migrate
• Update REMIND scenario from 3.5.0 to 3.5.2
• Fix error in inventories for PV farm transformer: the amount of copper was overestimated by a factor of 10
• Extended country-specific IEA Task 12 PV inventories to all countries

Fix bug when generating scenario reports

Full Changelog: v.2.3.2...v.2.3.3