Brightcon 2026, hackathon and courses

Highlights / Discussion points

Takeaway: ElecImpact provides a graph-based data structure that makes life-cycle inventory (LCI) data and models reusable, traceable, and directly interoperable with existing tools such as Brightway. The code lives in a multi-package monorepo, and the data lives in a separate repository.
Open modular tools: ElecImpact is fully open source under a permissive license, with a modular, extension-based design. It builds on existing formats, proposing for example an “augmented” Brightway format that preserves types and symbolic formulas.
Reproducibility by design: Products, processes, and models are typed nodes. Parameters and exchanges are symbolic formulas in a reusable formula graph, so any impact figure can be traced back to models and data and recomputed end-to-end.
Composable and scalable: Independent studies and their processes can be merged into one connected super graph, and alternative ways to assess a given impact are exposed. Design-of-experiments scenarios can then be propagated in memory through the graph and their impacts
computed as a single vectorized batch. This turns sweeps that would take months in a spreadsheet loop into minutes.
Community effort: A multi-laboratory GDR DEFIE collaboration (INSA Rennes, IRISA, CEA-List, Mitsubishi Electric R&D Centre Europe (MERCE), Fraunhofer IZM).


Concise description:

Context: Life Cycle Assessment increasingly underpins regulation and eco-design. Yet its life-cycle inventory (LCI) data and models are often opaque, hard to reuse, and difficult to maintain. This is especially pronounced for electronics and ICT, where impacts rest on composite, hierarchical, and only partially validated models [1]. Four requirements for trustworthy models then emerge [1]: explicit
lineage, well-defined scope, end-to-end traceability, and managed non-obsolescence. Credible practice therefore needs tooling for the collaborative construction of open, versioned, and traceable
inventories that integrate with existing LCA software.

Approach: ElecImpact is an open initiative of the GDR DEFIE (CNRS, https://defie-cnrs.fr/) that structures and shares LCI data and models through a fully graph-based architecture. Products,
processes, and models are typed nodes. Parameters and exchanges are symbolic formulas in a reusable formula graph rather than fixed numbers. Impact and parameter-conversion models attach to
nodes by reference, so results record which models produced them, and models can be updated without breaking downstream studies. Linked studies can also be merged into one connected super graph. The design is modular: a small, domain-agnostic Python core defines the LCI ontology and import/export. Anyone can add opt-in extensions (data models, parsers) without having to modify the core. For interoperability, formats such as an “augmented” Brightway format let inventories be exchanged with Brightway and other tools. For computation, vectorized batches are used both to
propagate parametric inventories in memory and to compute their environmental impacts (reusing Brightway functions), keeping large scenario sweeps tractable.

Brightcon spirit: Everything is open source and versioned, so results can be traced back to models and data and recomputed end-to-end. Extensions (Python packages) can be developed and
maintained outside the main repository. ElecImpact is a community effort and invites everyone to reuse, extend, and contribute.


Links

• Website: https://defie.frama.io/elecimpact_docs/
• Source code: https://framagit.org/defie/elecimpact
• Figure (graph-based architecture): https://defie.frama.io/elecimpact_docs/#paradigm-2-the-database-is-a-
  graph
• Figure (import/export pipeline): https://defie.frama.io/elecimpact_docs/concepts/technical_design_principles/

Reference
[1] V. Corlay, D. Bekri, M.-A. Lacroix, M. Pelcat, M. Péralta, P. -Y. Pichon, L. Saillenfest, O. Weppe, S. Rumley, 2026. All LCA models are wrong. Are some of them useful? Towards open computational LCA in ICT. In: Proc. Sustainable Computing Workshop 2026 (SCW ’26), Catania, Italy. ACM. https://arxiv.org/pdf/2604.06290