| title | tags | authors | affiliations | date | bibliography | ||||||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
pathways: life cycle assessment of energy transition scenarios |
|
|
|
24 May 2024 |
paper.bib |
pathways is a Python package that conducts Life Cycle Assessment (LCA) to evaluate
the environmental impacts of products, sectors, or transition scenarios over time.
Unlike most energy (ESM) or integrated assessment models (IAM), pathways offers
a clearer view on impacts caused by a scenario by considering supply chain relations
between producers and consumers, thereby addressing direct and indirect emissions.
Unlike the reported emissions in ESM and IAM scenarios, which focus primarily on operation,
pathways allows reporting the environmental impacts of infrastructure build-up
and decommissioning. Finally, scenarios can be characterized across a wide range of
indicators which are usually not included in ESM or IAM: land use, water consumption,
toxicity impacts, etc.
Most IAMs and ESMs project cost- or utility-optimized future scenarios within specified greenhouse gas emissions trajectories, outlining changes needed in regional energy mixes and means of transport for global warming mitigation [@Riahi:2017]. Prospective Life Cycle Assessment (pLCA) is crucial for evaluating the environmental performance of existing and emerging production systems, with a growing body of literature in scenario-based pLCA for emerging technologies [@Bisinella:2021].
Extending present-day life-cycle inventories into the future using IAM outputs,
initially explored by @MendozaBeltran:2018 and formalized by the Python library
premise [@Sacchi:2022], forms the methodological basis for pLCA. Efforts in pLCA
focus on improving forecasting accuracy. Performing system-wide LCAs with
adjusted life cycle inventories at each time step has potential to enhance
sustainability assessments, broadening focus beyond greenhouse gas emissions
to include broader environmental impacts like land use, water consumption,
and toxicity, addressing both direct and indirect emissions. However, system-wide
LCA remains challenging due to computational costs and methodological
complexities, such as defining functional units based on IAM outputs and
resolving double-counting issues [@Vandepaer:2020; @Volkart:2018].
Several studies characterize energy scenarios with LCA, including @Gibon:2015, @Rauner:2017 and @Pehl:2017, who quantified ESM or IAM scenario outputs using a hybrid-LCA framework. There is also the work of @Xu:2020, who developed the ambitious EAFESA framework aiming for bidirectional coupling between ESM and LCA. Yet, these studies focused on specific sectors or technologies and have not yet generalized to broader scenarios and indicators, nor have they made their implementations widely available.
Beyond conventional pLCA approaches, several tools and frameworks have been developed
that leverage LCA data to support further analysis, often through automation and integration
with broader modeling frameworks. For example, the ODYM-RECC framework integrates LCA data
to assess resource efficiency within climate mitigation scenarios, providing insights on
material demand and supply chain impacts [@RECC:2021]. Similarly, the Mat-dp tool, when
supplied with suitable input data, can be used to calculate materials needed and estimate
environmental impacts of transition scenarios [@Mat-dp:2022; @Mat-dp:2024]. However, because
these tools depend on exogeneous input data, they are not designed to
systematically consider the time-dependent technology mixes influencing the production system.
This limits their ability to endogenously and dynamically assess evolving environmental impacts
and material demand, restricting consistency with the scenario assessed.
To address these challenges, the open-source library pathways utilizes the
LCA framework brightway [@Mutel:2017] to systematically evaluate
environmental impacts of energy transition scenarios. pathways works with
data packages containing LCA matrices adjusted to each time step of the
ESM/IAM scenario, providing detailed and transparent insights into
scenario environmental impacts. pathways works particularly well with
data packages produced by premise, but can be used with any ESM/IAM scenarios
and LCA databases. Using LCA matrices which have been modified to reflect
the scenario's time-dependent technology mixes ensures a consistent and coherent
characterization of the scenario.
pathways reads a data package containing scenario data, mapping information,
and LCA matrices. The data package should be a zip file containing the following
files:
datapackage.json: a JSON file describing the contents of the data package- a
mappingfolder containing amapping.yamlfile that describes the mapping between the IAM scenario variables and the LCA datasets - an
inventoriesfolder containing the LCA matrices as CSV files for each time step - a
scenario_datafolder containing the scenario data as CSV files
pathways reads the scenario data files (1 in Figure 1), and iterates,
for each time step and region, through technologies with a non-null
production volume. For each technology, pathways retrieves the corresponding
LCI dataset by looking it up in the mapping file (2 in Figure 1). The lookup
indicates pathways which LCA matrices to fetch from the data package (3 in Figure 1).
The LCA matrices are loaded in bw2calc (the LCA calculation module of brightway)
and multiplied by the production volume (see 4 in Figure 1). The results are aggregated
and saved in a dataframe, where impacts are broken down per technology, region,
time step, geographical origin of impact, life-cycle stage and impact assessment
method (6 in Figure 1).
Some post-processing is done on the inventory matrices, including managing double counting. Double counting occurs when resource demands are counted multiple times across interconnected system components, inflating environmental impacts. This issue is particularly relevant when the reference scenario (e.g., from an IAM) already accounts for total regional demand, such as electricity or transport. For example, if electricity and steel production are interdependent, evaluating total electricity demand as defined by the scenario may lead to overlap: electricity requires steel, and steel production, in turn, requires additional electricity beyond the initial total. This overlap results in duplicative demand estimates.
To address this, the original LCI database is adjusted by zeroing out all regional energy inputs that the energy system model accounts for and might demand during the system's life cycle, following the same workflow presented in @Volkart:2018 (see 5 in Figure 1). Practitioners are required to selectively cancel out overlapping activities already accounted for by the scenario. We use a modular approach in this adjustment process, where practitioners, based on their understanding of the model generating the scenario, can select specific activity categories (e.g., electricity, heat, or specific product inputs) to exclude. For instance, if the IAM models regional electricity generation, the corresponding electricity inputs in the LCA system for upstream processes are removed to prevent double counting. Returning to the electricity-steel example, this means the practitioner would exclude electricity inputs for steel production within the LCA, as the scenario’s total electricity demand already covers this requirement.
This process is implemented in the remove_double_accounting function, which modifies the
technosphere matrix to remove redundant entries. The function identifies flagged
products for removal, locates the associated rows, and zeroes out the corresponding positions
taking any specified exceptions. For instance, in the electricity-steel example, the function
would find the row corresponding to regional electricity and cancel out the input in the column
associated with steel production, effectively preventing double counting of electricity demand.
This modular approach enhances transparency and traceability, making it easier to document and
track which system components are modified, ensuring consistency between the scenario outputs and the LCA.
Finally, Global Sensitivity Analysis (GSA) can be performed on the results.
Currently, pathways supports the use of the SALib library for GSA [@Herman2017; @Iwanaga2022],
notably the Delta Moment-Independent Measure (DMIM) method [@BORGONOVO2007771], to rank
the influence of the database exchanges on the results.
A detailed example notebook
is available for using pathways with a sample data package.
By systematically updating and integrating LCA matrices over time, pathways improves the accuracy and relevance of
environmental impact assessments for transition scenarios. This tool fosters greater alignment between LCAs and ESM/IAM
outputs, enhancing the consistency and reliability of environmental assessments across different modelling platforms.
Additionally, pathways offers a detailed and structured workflow that enables IAM
modellers to incorporate LCA into their analyses. This opens new avenues for these modellers to enhance the
environmental dimension of their work.
Designed to be both reproducible and transparent, pathways facilitates collaboration and verification within the
scientific community. This approach ensures that improvements in environmental impact assessments are accessible and
beneficial to a broader range of stakeholders.
pathways is a tool that evaluates the environmental impacts of transition
scenarios over time using time-adjusted and scenario-based LCA matrices. This
approach allows for characterizing the environmental impacts of a scenario
across a wide range of indicators, including land use, water consumption,
toxicity impacts, etc. It also allows to attribute supply chain emissions
to the final energy carriers, thus providing a more detailed and transparent
view of the environmental impacts of a scenario.
The authors gratefully acknowledge the financial support from the Swiss State
Secretariat for Education, Research and Innovation (SERI), under the Horizon
Europe project PRISMA (grant agreement no. 101081604). The authors also thank the
Swiss Federal Office of Energy (SFOE) for the support of pathways via the SWEET
program for the consortium SURE.