Document Matrix Dimensions (and maybe Build Pipeline?)

[michaelweinold]

Request Category

Changes to existing page

Page Title

Theory > Brightway Structure

Description

I expect a characterization matrix to be of dimension [# impact categories x # biosphere flows]. See, for instance Section 8.4.1. in Heijungs & Suh (2002).

However, inspecting the characterization matrix of a Brightway LCA class, which was instantiated with a single impact assessment category gives something completely different:

lca = bc.LCA(
    demand={bd.Database("ecoinvent-3.10-apos").random():1},
    method=('IPCC 2021', 'climate change', 'global warming potential (GWP100)')
)
lca.load_lci_data()
lca.load_lcia_data()

Inspecting the method, we see that it has 184 associated characterisation factors:

bd.Method(('IPCC 2021', 'climate change', 'global warming potential (GWP100)')).metadata

{'description': '',
 'filename': 'LCIA_Implementation_3.9.xlsx',
 'unit': 'kg CO2-Eq',
 'abbreviation': 'ipcc-2021cg.7f3bec1f3f63e5211c9815c52a5f928c',
 'num_cfs': 184,
 'geocollections': ['world']}

But the resulting matrix has this dimension:

lca.characterization_matrix.todense().shape

(2647, 2647)

This does not even match the number of biosphere flows:

len(bd.Database('ecoinvent-3.10-biosphere'))

4362

From what I can see, the LCA.characterization_matrix is build in this pipeline of function calls:

bw2calc.lca.LCA > bw2data.compat.prepare_lca_inputs > bw2calc.lca.LCA.load_lcia_data > matrix_utils.mapped_matrix

At what point we go to a dimension of 2647x2647 - I can't tell.

Can you help?

• I could help with the content of this documentation request.
• I could help with the technical implementation of this documentation request.