carculator_bus

Prospective environmental and economic life cycle assessment of city buses and coaches, including depot, opportunity and in-motion charging.

Developed at the Paul Scherrer Institute. This checkout prepares the unpublished release 0.1.2; see CHANGELOG.md for release status and changes.

Installation

Use Python 3.12 (>=3.12,<3.13) and a fresh environment. The shared runtime requires NumPy >=1.26.4,<2.

python3.12 -m venv .venv
source .venv/bin/activate

On Windows, activate with .venv\Scripts\activate. After publication, install this release from PyPI:

python -m pip install "carculator_bus==0.1.2"

Before publication, use the matching source checkouts as described under development. Core calculations use bundled resources and need no Brightway project, ecoinvent installation or network access. The matching carculator_utils runtime installs brightpath>=1.0.0a6,<1.1 for Brightway Excel, SimaPro CSV and openLCA JSON-LD exports. It uses Brightpath's v1 API, currently an alpha; model and LCIA calculations do not import Brightpath or Brightway.

The excel extra remains a compatibility alias. To select the tested legacy Brightway stack (bw2io<0.9, bw2data<4, bw2calc<2), use:

python -m pip install "carculator_bus[brightway]==0.1.2"

Brightpath also installs bw2io without that extra. Export targets ecoinvent 3.12 cut-off by default; legacy 3.9/3.10 targets reject suppliers without verified older links. External suppliers must be matched to the corresponding background in the destination tool. See inventory export.

Quick start

from carculator_bus import (
    BusInputParameters,
    BusModel,
    InventoryBus,
    fill_xarray_from_input_parameters,
)

inputs = BusInputParameters()
inputs.static()
_, array = fill_xarray_from_input_parameters(
    inputs,
    scope={"size": ["13m-city"], "powertrain": ["ICEV-d", "BEV-depot"], "year": [2025]},
)
model = BusModel(array)
model.set_all()
print(model["TtW energy"])  # kJ per vehicle-kilometre

inventory = InventoryBus(model, functional_unit="pkm")
impacts = inventory.calculate_impacts()
print(impacts.sel(impact_category="climate change").sum("impact"))

Bus model cost outputs are already per passenger-kilometre; inventory normalization is a separate choice.

Inventory export

With the completed inventory from the quick start:

workbook = inventory.export_lci(software="brightway2", format="file", directory="exports")
simapro_csv = inventory.export_lci(software="simapro", format="file", directory="exports")
foreground_zip = inventory.export_lci(software="openlca", format="file", directory="exports")

Retain exactly one sample before constructing the model and inventory. Each year gets its own export; multiple years return a list. Exports preserve the original inventory and calculated impacts. The default export_lci() returns an unlinked Brightway importer.

The openLCA ZIP contains foreground processes, without an ecoinvent background or LCIA methods; map external providers and elementary flows before calculation. SimaPro CSV uses Latin-1 and warns when omitting custom noise flows. See the export guide for return types, sample selection and format-specific limitations.

Modelling and validation

For a guided explanation of the inputs, units and assumptions, read Understanding inputs and results. The validation examples include bar charts, source records and the limits of each comparison. Measured energy comparisons are separate from the model-to-model comparison of the background database update.

Battery unit prices supplied by users now survive chemistry selection and cost adjustment. Use battery_costs for explicit prices scoped by vehicle, year and sample; see battery-cost inputs.

The vehicle models include tabulated 2025 parameters and documented temporal extensions. These combine engineering assumptions and selected calibration evidence; they are not independent measurements for every vehicle configuration.

TtW energy is in kJ/km. For BEVs it is net stored-energy depletion; model.battery_terminal_energy reports terminal DC separately, while electricity consumption is grid electricity in kWh/km. Identify the measurement boundary before comparing energy outputs. Availability-masked zeroes do not represent physically zero consumption.

Charger-equipped buses deliberately include at least one replacement battery over service life, even if cycling demand alone implies none. The initial pack is additional. This is a fleet-life assumption, not a calibrated ageing model; see battery replacement accounting.

Supported background scenarios are SSP2-NPi, SSP2-PkBudg1000, SSP2-PkBudg650, and static. ReCiPe supports midpoint/endpoint and EF midpoint. Use fresh model instances for independent cases. inputs.stochastic(n, seed=...) seeds parameter draws and the projected-cost factors used by the current array builder. Keep its auxiliary sample coordinates; older hand-built arrays do not recover that seed automatically.

See validation and limitations, migration notes and the documentation.

Development and release

Use matching sibling checkouts, especially carculator_utils 1.3.7 or newer:

python -m pip install -e "../carculator_utils[test,excel,brightway]" -e ".[test,docs,excel,brightway]"
python -m pip check
python -m pytest
python -m sphinx -b html docs docs/_build/html

The docs extra includes the extensions used by this repository. See RELEASING.md for artifact verification, release order and publication.

Support and license

Contact carculator@psi.ch or open an issue. Maintained by Romain Sacchi, with contributions from the carculator development team. See contributing. Licensed under BSD-3-Clause.

The shared LCA background now uses premise 2.5.4 and ecoinvent 3.12 cutoff. The former NMC-523 option is replaced by NMC-532 (actual Ni:Mn:Co 5:3:2 inventory); update custom chemistry selections. See docs/release.rst.