Page 290 - Materials Chemistry, Second Edition

276 R.K. Rosenbaum et al.

In LCA practice, the terms variability and uncertainty are often not distinguished
or overarching one another (i.e. variability is often included as one aspect of
uncertainty). However, for their important differences described before, it is rec-
ommendable and good practice to quantify and maintain both well separated as this
will allow us to put this information to good use when interpreting and improving
LCA results. We will come back to that later.
The sensitivity of a model describes the extent to which the variation of an input
parameter or a choice (e.g. time horizon in the functional unit) leads to variation of
the model result. A model is sensitive toward a parameter if a small change in this
parameter will result in a large change in the model result, whereas a model is
insensitive toward a parameter if any change in this parameter will have no (or
negligible) effect on the model result (which in certain cases might indicate that this
parameter may not be needed in the model, or at least that it is not an important
input parameter for this particular value of the model result). Sensitivity may be
analysed for both continuous and discrete input parameters, and it can also be
analysed for choices leading to discrete sets of input values. For example, the
choice of LCIA method is always a discrete choice between a certain number of
fixed options (i.e. available methods). It is worth noting that the term sensitivity is
used in various and inconsistent ways throughout literature and no agreement on its
exact definition exists. Two main uses could be distinguished: (1) For some authors
sensitivity includes the effect of uncertainty and thus considers the range of vari-
ation of input parameters as a function of their uncertainty (which hence needs to be
known), varying them all at the same time. This is also called global sensitivity
analysis and is essentially what this chapter refers to as uncertainty analysis.
(2) Others define sensitivity solely as the effect of a certain change in input on the
output applying a predefined variation without considering the uncertainty. This is
analysed by varying one parameter at a time and also called local sensitivity
analysis. In the context of this book and many publications in the LCA community,
sensitivity only describes the variation of a result due to variation of an input or
choice, without considering its uncertainty, i.e. local sensitivity.

11.2.2 Defining Accuracy and Precision in the LCA Context

When talking about uncertainty, a number of terms are often used in conjunction or
interchangeably which seem to be synonyms but in fact are not. Two such terms are
accuracy and precision. The definition of these terms in general English dictionaries
varies to some extent, the Oxford English Dictionary for example defines accuracy
as technical noun being “The degree to which the result of a measurement, cal-
culation, or specification conforms to the correct value or a standard” and precision
as technical noun being “Refinement in a measurement, calculation, or specifica-
tion…”. Therefore, both terms are independent and while accuracy refers to the
correctness of a value, precision relates to the relationship among multiple mea-
surements or calculation results. It is therefore useful to have a closer look at the