Page 49 - 2417

Page 2 of 22 François et al. Carbon Footprints 2026, 5, 22

method ignores compatibility constraints between static and dynamic approaches and because it does not
differentiate delay factors for distinct GHGs. However, delaying emissions reduces GWP but still raises GTP,
questioning the use of GWP as a dynamic indicator, as it may falsely suggest a declining impact when it is actually
increasing.

INTRODUCTION
Dynamic Life Cycle Assessment (LCA) covers the notion of taking temporality explicitly into account in an
LCA study. More precisely, a dynamic LCA is defined as the combination of two models: (i) a dynamic life
cycle inventory (LCI), modelling time-dependent processes in the technosphere; and (ii) a dynamic life cycle
impact assessment (LCIA), representing the time-varying nature of the expected effects within the
ecosphere . Dynamic LCA is not specific to climate change, but it has been initially developed concerning
[1]
that specific impact category [2,3] . Each impact category would require a very different dynamic
characterization approach, depending on the underlying impacted natural phenomena. However, only few
studies jointly implement both a dynamic inventory and a dynamic climate-change characterization, despite
the methodological need for doing so .
[4]
LCA approaches regarding climate change assessments have been classified into four categories for time
accounting : (1) Purely static approaches, with no consideration for time; (2) Partially dynamic approaches
[5]
combining a static inventory and an LCIA with credit for carbon storage; (3) Partially dynamic approaches,
where a dynamic inventory is assessed through a static LCIA method, and (4) Fully dynamic approaches
based on both a dynamic LCI and LCIA. For the sake of clarity, instead of using these categories as defined
by , we prefer to refer in the remainder of this article to static, partially dynamic and fully dynamic
[5]
approaches, as illustrated in Figure 1. A partially dynamic approach is characterized by a temporally explicit
foreground LCI, whereas a fully dynamic approach involves a temporally explicit inventory for both the
foreground and background LCIs. In either case, the LCIA is temporally explicit.

For a fully dynamic approach, a basic understanding of the chronological progression of life cycle processes
in the foreground system and/or the temporal equations is insufficient. It is also essential to be able to
temporally align all background system processes with respect to those in the foreground system, i.e. “to
combine the different dynamics of the unit processes together” . An example of this issue is provided in
[6]
Figure 1 for the case of a building. It is required to trace back in time the production of each composing
material, and then further in time their use and end-of-life flows relative to the construction timeline.
Similarly, each intermediate flow required by the production processes of these composing materials has to
be traced in time relatively to the production, use and end-of-life times of this specific material, etc. Such an
approach requires a mathematical framework and a computing method able to recreate a chronological
database from the temporal description of a product life cycle. Additionally, temporal LCI models can vary
in complexity, ranging from basic chronological ordering of processes to more complex dynamic
relationships. In the latter case, the value of an inventory flow at a given time t is a function of its value at the
preceding time step, as exemplified by models of biomass growth in silviculture. Two types of approaches,
grounded in distinct mathematical techniques, have been put forward to address this issue: the first relies on
the method of limited expansion , while the second utilizes dynamic graph search algorithm . These have
[7]
[6]
later been improved and operationalized [8-10] . However, these fully dynamic methods can only be
implemented on specific databases for which all the intermediate flows are available (such as unit process
models in the ecoinvent database). Their implementation also demands a high level of programming
expertise.