Streamlining of Sensitivity Analysis

[bsteubing]

Context:

Many LCA studies include Sensitivity Analyses. Most of them do OAT (one at a time) SA. This can take two forms:

• varying a parameter by +/- a percentage
• varying a parameter within known/possible bounds, e.g. from 0.8-0.9

Idea

The AB could streamline this. For example via the import of a "parameter-sensitivity" or "exchange-sensitivity" file. This would be organized as the "parameter-scenarios" or "scenario difference file" format, but with a simplification. Instead of having to develop a column for each "scenario", the file would just need 2 columns: min and max. These would contain the bounds (+/- x% or the absolute values). From this, all sensitivity analyses could be conducted "one at a time".

To consider / think about:

Where to implement this?

• either as part of "scenario LCA", with an option "OAT sensitivity file", or
• as part of a new "sensitivity LCA" mode (but that might be an overkill for now; and if, then the question is if GSA should also move to that mode)

How to combine this with "scenario LCA" / superstructure approach? Which results should be shown? If the additional implementation effort is worth it given that this can already be done via "scenario LCA"