Drawing from Semantic Catalogs for Life Cycle Assessment Data, the LCA data model has three distinct entity types:
- Flows are "things in the world" that can be observed and described in quantitative terms. I regard the unique concept of the flow as the definitive conceptual class of LCA and industrial ecology generally. In our implementation, every flow must be assigned a specific reference quantity to describe its extent, with the default being "number of items".1
- Quantities are extensive properties that can be used to describe the extent or dimensionality of specific, observed flows. Mass and price are quantities; LCIA indicators are also quantities.
- Activities are interactions of several flows exchanged across a well-defined spatial and temporal boundary. I use the term "Process" to describe a concrete inventory model of a specific activity, i.e. an enumeration and quantification of all flows that are co-exchanged across a boundary having a specific spatiotemporal scope.
1- Note: although the "Semantic Catalogs" article describes flows as having mandatory Compartments, this requirement has been abandoned in favor of contexts.
Additionally, there are ancillary concepts that go along with these primary data types:
- Units of measure. Several different units can be used to measure the same quantity, and the same unit can also be used to measure different quantities. A unit is simply a string, and can indicate a measurement but is insufficient to identify a quantity. For clarity, I like to create distinct indicator units for each quantity.
- Contexts are as defined in Critical review of elementary flows in LCA data, namely "...categories typically describing an environmental context of the flow origin or destination ... that include[s] the flow directionality." An exclusive nomenclature of Contexts (but including synonyms) is integral to LCIA; however, each data source is free to natively define any contexts, which can then be harmonized with the LCIA engine. A context is always associated with an exchange and not with a flow entity. Similarlly, there is no such thing as an "elementary flow", only an elementary exchange.
- Directions describe the transit of a flow across a boundary. Directions can have only two values, "Input" and "Output", which can be most efficiently represented in python using canonical strings.
For convenience, clusters of entities can be related together into hybrid objects called "exchanges" and "characterizations":
- An exchange comprises a process, a flow (with reference quantity), a direction (with respect to the process), and an optional termination (designating the remote end of the exchange, either another process or a context)
- A characterization comprises a flow (with reference quantity), a characterized quantity, a context (with an inherent direction), and an optional spatiotemporal scope.
Using these objects, we can define the two quantitative relations that collectively describe the full range of computations that occur in life cycle assessment, the exchange relation and the quantity relation:
-
Given an exchange, the exchange relation is the magnitude of the exchanged flow (in terms of its reference quantity) that is associated with a unit of a specified reference flow;
-
Given a characterization, the quantity relation is the magnitude of the characterized quantity that is equivalent to a unit of the characterized flow's reference quantity.
Both relations map n-tuples of entities to real numbers. The exchange relation is used to compute the values of nonzero entries in the A and B matrices; the quantity relation is used for unit conversions and to compute the values of nonzero entries in the C matrix.
I first used the term "life cycle fragment" in my work on publishing the CalRecycle Used Oil LCA model. A fragment refers to a connected subgraph of a larger LCA foreground model, explicitly excluding aggregated background activities, which can be aggregated to a single node through a traversal procedure. Conceptually, they are similar to "plans" used as modeling components in GaBi. I tend to use the term "fragment" rather sloppily to refer to both a subgraph, and also to an individual link within the subgraph, better called a "fragment link."
A fragment link can be thought of as a parameterized exchange that belongs to the foreground of a model. It has five components:
| Component | Optional? | Data type | Comment |
|---|---|---|---|
| Parent node | Yes | Fragment or None |
The vantage point from which the exchange is observed |
| Flow | No | Flow entity | The material / service / substance that is observed |
| Direction | No | Input or Output |
with respect to the parent |
| Exchange value | Yes | Float / default 1.0 | Parameterizable |
| Termination | Yes | None, self, sub-fragment, process, or context |
Parameterizable |
A fragment link's termination determines how the fragment link works in a disclosure. Fragment links with no terminations are "cutoffs" that operate either as un-met requirements or co-products of the fragment to which they belong.
Fragments that are interior parts of a foreground model are typically terminated to themselves (the link corresponds to both a nonzero entry in the Af matrix and to a particular row and column of that matrix. These links are called foreground fragments.
Fragment links can also terminate to processes, meaning that the fragment's flow matches up with one of the reference exchanges of the process. The process is then included in the foreground of the model, and the process's other exchanges can be used to define child fragment links.
Finally, fragment links can terminate to other fragments, in which case the termination target is called a "subfragment" and is traversed separately. The outcome of the subfragment traversal is then used to assign the exchange values to any child links of the parent fragment.
A collection of fragment links is well-defined if every fragment link's parent node is either None or in the collection. Such a collection forms a directed forest. Every fragment link with None parent is called a "reference fragment", or simply a fragment by the above definition. The set of reference fragments equals the distinct set of product systems modeled by the forest.
Please see:
- Repository README
- antelope.md for a discussion about the features and capabilities of the Antelope interface and Antelope Catalog
- antelope-interface.md for a semi-complete description of the Antelope interface