{% extends "base.html" %} {% block body %}
Matrix graphs allow an easy visual inspection of database structure.
Each dataset should produce its own reference product. Failing to produce your own reference product could result in a singular technosphere matrix, making LCA calculations impossible.
Multioutput processes can be used in Brightway2 (see Multioutput processes in LCA), but should be handled with care. Each multioutput process should be inspected carefully to make sure the intended effects will occur.
{% if mo %}There are no multioutput processes in this database
{% endif %}This are the 20 most important processes, when the technosphere matrix is analyzed using the Page Rank algorithm. Page Rank scores are normalized to the number of processes in the database, i.e. each score is actually page rank score * number of processes in database. Anything higher 50 is pretty high - these processes should be audited for completeness and accuracy, as they will be very important in calculating impact assessment scores.
In general, each exchange should represent a unique pairing of an input process and amount. Duplication of exchanges makes uncertainty and sensitivity tests harder, and are an easy path for introduced errors when some exchanges are updated and others are not.
Exchange type | # Exchanges | # Unique exchanges |
---|---|---|
Technosphere | {{ ue[0] }} | {{ ue[1] }} |
Biosphere | {{ ue[2] }} | {{ ue[3] }} |
The LCA community has an uneasy relationship with uncertainty, as most practitioners are not statistical experts. Errors in the uncertainty distributions that are included in most databases occur too often. Below is a table of the uncertainty distributions used in this database, and the number of errors found.
Uncertainty type | # Exchanges | # Errors |
---|---|---|
{{ line[0] }} | {{ line[1] }} | {{ line[2] }} |
Processes which consume some of their own reference product as an input can introduce subtle math errors, as there are at least two different ways of thinking about the math of these datasets (see Two potential points of confusion in LCA math). Brightway2, the ecoinvent centre, and SimaPro all use the gross production assumption, where the activity produces the production amount minus the amount consumed as an input, e.g. 1 kilogram of production with 0.5 kilograms of reference product would mean that the activity only produced 1 - 0.5 = 0.5 kilograms, and 0.5 would be the value put in the technosphere matrix.
These process datasets should be carefully checked to make sure that the gross production assumption is used consistently.
System processes are "black boxes", where the dataset links only to biosphere flows, and no supply chain (i.e. no technosphere exchanges) is present. They are highly discouraged because they are impossible to understand, analyze, or update.
Datasets with more than 500 technosphere inputs are discouraged.