Opening

Battery circular-economy governance.

A meta-synthesis of 25 papers. Battery circularity does not hinge on technology — it hinges on how firms, regulators, platforms, communities, and infrastructure coordinate value.

86Canonical flows
363Evidence rows
25Papers
99%Evidenced
86%Well-triangulated

What this site does

Five governance archetypes; twenty-two recurring actors; 86 evidenced flow patterns extracted from the case literature. Scroll for the diagnosis, the recurring cast, the governance space, and the five archetypes; open any archetype's drawer for the underlying quotes.

Continue
The diagnosis

Engineered to last.
Killed by their ecosystem.

Lithium-ion cells routinely outlast the products built around them. So why does almost no industrial pathway exist to put them back to work? The technology isn't the bottleneck. The ecosystem is.

Four structural failures recur across the literature. Together they amount to a system that destroys value it has every reason to capture.

Failure 01

Cells outlast packs

Housings and electronics fail before the cells. Scrapping the pack scraps still-usable cells with it.

Failure 02

Designs block repair

Welded modules and encrypted firmware lock independent repairers out of the packs they encounter.

Failure 03

Data sits in silos

Provenance, charge history, and state-of-health data exist, but never reach the actors who need them.

Failure 04

Rules outpace capacity

The EU Battery Regulation tightens obligations faster than the infrastructure to discharge them comes online.

Continue
A · The recurring cast — minimum viable model

The cast recurs.
The script changes.

Twenty-two actors and 86 coded flow patterns make up the full canvas. Across the five archetypes that follow, eleven actors and a small set of structural arrows recur — the minimum viable model. What changes archetype to archetype is which flows activate, and who runs the table.

The orange-red cluster — Refurbisher, Recycler, Repairer — is the structural backbone. Three actors doing materially different things (rebuild to spec, reduce to elements, fix on the spot) but bound by the same governance question: who feeds them, who diagnoses, who buys.

Two arrows close the cycle: the Battery recycler returning material back to Smelter and to the Battery cell mfr. Strip those out and the diagram is a tree, not a circular economy. The Grey-market reseller remains the structural rival to every legitimate path — co-destruction and leakage are coded as empirical valence on otherwise ordinary value-type arcs.

What separates the five archetypes that follow is not who is at the table. It is who is running it. Each archetype reuses this same MVM and projects a different dominant battery path through it.

How was this map derived?

The MVM is computed deterministically from the five archetype canonical loops. Each archetype carries a hand-curated canonical_loop — the smallest set of source → target edges whose presence the archetype's governance theory requires for the loop to close as a circular economy. From those five sets, the MVM is the algorithmic union:

  • MVM actors = union over archetypes of {actor ids touched by canonical loop} → 15 actors
  • MVM edges = union over archetypes of {canonical (from, to) pairs} → 25 canonical flows

Each edge is then audited against the workbook: 19 are evidenced (solid Atlantic blue) and 6 are coding gaps (dashed grey ghost edges). The seven actors excluded from the MVM — Mining company, Grey-market reseller, Reverse logistics, Independent repairer, Industry alliance, Research institute, Infrastructure orchestrator — exist in the wider ecosystem coding but do not feature in any archetype's canonical loop. Their absence is itself a finding: the structural backbone of battery circularity sits in 15 of the 22 coded roles.

Positions are hand-curated for legibility. The forward value chain runs clockwise from OEM at top, through 1L sale (Vehicle dealer, EV owner), data routing (Diagnostics provider, Orchestrator) and 2L manufacturing (Refurbisher), to the 2L market (Installer, 2L customer) at the bottom. The reverse loop returns counter-clockwise: NGO and governance actors on the left, reverse logistics (Collector, Recycler, Smelter) on the upper left, closing materials back to Cell mfr and then to OEM.

hover an actor or flow for details · drag nodes to reposition
Canonical edge · evidenced
Canonical edge · coding gap
Minimum viable model · 15 actors · 25 canonical flows (19 evidenced, 6 coding gaps). Hover any underlined actor in the text to highlight it here.
Continue
A1 · The governance space

Five archetypes,
five distinct regions.

25 papers inform this synthesis. Bubble area scales with coded flow rows where each archetype is the dominant pattern. The axes — who orchestrates and what value organises the loop — locate each archetype in the working space.

X-axis · who orchestrates. From a single firm running the loop end-to-end, to a network of independent actors coordinated by data infrastructure or mutual interest.

Y-axis · what value organises the loop. From private commercial value (revenue, residual value, supply security) to institutional public value (compliance, legitimacy, social access).

Bubble area scales with coded flow rows in each lens. Industrial symbiosis dominates the workbook (53 rows); Regulation-driven appears in modelled form only — no edge in the current evidence base has Regulation-driven as its dominant pattern.

Empty top-left quadrant. Institutional and firm-hierarchical is where state-owned-enterprise or industrial-policy-driven loops would live. The current sample includes none.

Producer-led occupies the bottom-left because the OEM both orchestrates centrally and organises around commercial logic. Industrial symbiosis sits to its right: still commercial, but distributed across firms in physical proximity. Regulation-driven rises on the y-axis because it is institutionally framed. Community access rises and shifts right because it is institutionally framed and distributed across donors, NGOs, and civil-society actors.

Continue · read each archetype
A2.1 · Archetype 1 of 5 · Producer-led

Producer-led

The OEM's closed loop · A Volvo XC40 Recharge, year seven.

The owner trades the car in. A clause in the lease returns the battery (still at 78% state of health) from the Vehicle dealer to a Volvo-controlled remanufacturing facility, where technicians using Battery cell mfr. instructions replace eight modules and re-certify it. It ships back out with a Volvo warranty into the next XC40.

One firm owns every node. The loop closes because nothing leaves the Volvo perimeter.

Tier-1 suppliers feed material in. Battery recyclers take what is truly dead. A PRO (compliance) collects the fee. But the data, the residual value, and the customer all stay with the EV/auto OEM.

Why it works: the OEM has the strongest information advantage and the strongest brand incentive. Why it strains: it locks independents out of the second-life value pool. That is where the next archetype begins.

12 edges have this archetype as their dominant pattern in the workbook.

hover an actor or flow for details · open the evidence drawer to see the dominant battery path
Dominant path · evidenced Dominant path · coding gap Other MVM flows
Continue · next archetype
A2.2 · Archetype 2 of 5 · Platform-coordinated

Platform-coordinated

Routed by its passport · A Renault Zoe in Lyon.

The car's BMS uploads its final state-of-health to a third-party platform like Circulor, or a service built on the EU Digital Product Passport. The platform doesn't own the battery. It owns the routing.

Money flows backward; data flows forward.

A Diagnostics provider posts a verified SoH score. A Battery refurbisher bids for the pack on the platform's marketplace. Hours later, modules are listed for sale; a small home-storage installer in Sweden picks them up. Before the sale, diagnostics supplies a compliance attestation to the buyer's PRO (compliance). The Renault EV/auto OEM generated the original telematics, but the Ecosystem orchestrator is in control of the path.

Why it works: verified data lets actors who never met clear deals at scale, opening the loop to fleets of any size and OEM-agnostic refurbishers. Why it strains: if the data infrastructure isn't trustworthy and interoperable, the routing collapses and the loop reverts to bilateral OEM control.

18 edges have this archetype as their dominant pattern in the workbook.

hover an actor or flow for details · open the evidence drawer to see the dominant battery path
Dominant path · evidenced Dominant path · coding gap Other MVM flows
Continue · next archetype
A2.3 · Archetype 3 of 5 · Regulation-driven

Regulation-driven

The compliance chain · A German fleet pack via GRS Batterien.

Every producer that places a battery on the market is legally obliged to take it back. They don't do this individually. They pay a fee to a PRO (compliance) like GRS Batterien, which discharges the obligation across hundreds of producers.

Materials move. But the binding logic is documentary.

A fleet pack first becomes an obligation on paper. The PRO (compliance) converts that obligation into contracts, audits, and take-back capacity; a certified EoL collector books the pack into the official channel; a certified Battery recycler recovers cobalt, nickel, and lithium; and evidence travels back as tonnage data before becoming a compliance report to Government. The rival is the Grey-market reseller — who offers a simpler bargain: take the pack, move it offshore, capture the metal, avoid the fee. That is why Civil society / NGO is not merely watching from outside; it watches whether the regulated route is credible enough to beat the free-rider route.

Why it works: legal obligation aggregates compliance volume that no individual producer would generate alone, and routes it through a single accountable intermediary. Why it strains: if audit teeth and price signals don't make the legitimate route cheaper than the grey path, the official chain becomes a venue for the diligent minority.

This archetype is currently modelled rather than evidenced — no edge in the v2 workbook has Regulation-driven as its dominant pattern. The evidence below is drawn from edges where Regulation-driven has secondary weight, and from external/flow-level evidence that describes regulatory dynamics.

hover an actor or flow for details · open the evidence drawer to see the dominant battery path
Dominant path · evidenced Dominant path · coding gap Other MVM flows
Continue · next archetype
A2.4 · Archetype 4 of 5 · Industrial symbiosis

Industrial symbiosis

Five firms, five kilometres · Harjavalta industrial park, Finland.

Four firms inside a five-kilometre radius: a nickel smelter, a cobalt-sulphate plant, a precursor maker, and a planned cell-pack assembler. Add a Battery recycler at the edge and you have a closed loop made of pipes and short-haul trucks rather than ledgers.

No single dominant actor. Just geography, and shared infrastructure.

A used pack arrives at the recycler. Black mass goes three kilometres by truck to the Smelter & refiner, which pipes battery-grade nickel and cobalt sulphate next door to the precursor plant. The precursor plant reformulates active material and sends it to the Battery cell mfr.. Reverse logistics is shared across all five firms; knowledge from the Diagnostics provider travels openly because firms are not competing for the same downstream customer. The Infrastructure orchestrator who designed the park keeps the connections wired.

Why it works: geography and shared infrastructure substitute for contracts. Material throughput, second-life value capture, and demand pull all happen in a few square kilometres. Why it strains: the model depends on regional industrial policy and patient capital. Without the park, the same firms scattered across Europe would never close anything.

53 edges have this archetype as their dominant pattern in the workbook.

hover an actor or flow for details · open the evidence drawer to see the dominant battery path
Dominant path · evidenced Dominant path · coding gap Other MVM flows
Continue · next archetype
A2.5 · Archetype 5 of 5 · Community access

Community access

One battery, five solar pumps · Lake Victoria, Tanzania.

A bicycle-battery wholesaler in Ireland processes returned ebike packs. Most cells still hold 70% of their original capacity — too tired for fast cycling on a hill, more than enough to power a borehole pump or run lights at night. A small Battery refurbisher rebuilds modules and ships them to a partner organisation in rural Tanzania.

What is exchanged is mostly societal value — not strategic competitive advantage.

A community energy provider — JUMEME, supported by a public-private partnership — installs the modules into solar-water systems for fishing villages. An Independent repairer trains young technicians in the village to swap cells. An Installer / maintainer commissions the install and maintains it. A Civil society / NGO documents the impact and reports it to international donors and Government for subsidies. Disvalue threats come from cheap Grey-market reseller modules of unknown provenance dumped into the same markets.

Why it works: demand pull comes from a real, urgent welfare need; the model serves people no other archetype reaches. Why it strains: without subsidy, the economics do not work; the model is the most explicitly normative of the five, and that is the point.

2 edges have this archetype as their dominant pattern in the workbook.

hover an actor or flow for details · open the evidence drawer to see the dominant battery path
Dominant path · evidenced Dominant path · coding gap Other MVM flows
Continue · back to top
§ Dashboard

Explore the network yourself.

The full graph — twenty-two actors and eighty-six evidenced flows. Switch the governance lens to see which actors and which edges activate under each pattern. Toggle the loop, the backbone, flow labels, secondary-actor flows, and co-destruction highlights. Filter by ecosystem group or flow type. Hover any actor or flow for details.

Governance lens

The graph shows the full coded picture: every node is an actor that appears in at least one paper, every edge a flow with at least one verbatim quote. Hover any node for details; click "Show the evidence" to open the drawer for the selected archetype.

§ Methodology

How this map was derived.

The synthesis sits on 25 case-study and expert-survey papers, projected onto a shared three-layer ontology — 22 actors, 9 value-flow types in 3 families, 5 governance archetypes — through a five-stage pipeline. Every edge claim on this page traces back to a verbatim quote on a numbered page; non-edge claims are bracketed as interpretation rather than evidence.

  1. Corpus. 25 papers were selected from the IS-and-supply-chain literature on EV battery circular-economy governance: multi-case comparative designs, single-case deep dives, and expert-survey papers. The selection is documented in Battery_ecosystem_v2.xlsx · Papers sheet.
  2. Ontology. A shared lattice of 22 actor types, 9 value-flow types (Compliance · Data & informational · Economic & financial · Environmental · Operational & efficiency · Product & technical · Reputational · Societal · Strategic & competitive), and 5 governance archetypes is defined before any LLM coding begins. It is the shared vocabulary the 25 different lenses project onto.
  3. LLM coding. Each paper passes through a two-step prompt: identify the source and target actors from the 22-actor taxonomy; then identify the value-flow type, valence (positive / negative / neutral / mixed), and speech act (empirical / normative / background). Every coded claim returns a verbatim quote and a confidence rating; low-confidence codings (< 0.6) trigger a clarification follow-up.
  4. Edge enumeration. Coded quotes are joined into 421 source-target hypothesis pairs; 86 of these turn out to be evidenced active edges. Each edge accumulates a list of supporting evidence rows and supporting papers, surfacing triangulation: 99% of active edges have ≥1 evidence row, 86% have ≥2 independent papers.
  5. Propagation. Each paper's case-level archetype score (computed from two governance dimensions: who orchestrates and what value organises the loop) enters a weighted average at the edge level. The result is an archetype dominance score per edge — what colours the bubble chart and the subgraphs above. A documented limitation: papers about industrial parks that mention regulation as backdrop contribute to symbiosis-weighted edges, producing the paper-level averaging artefact (see §7.3 of the methods chapter).
2026-05-24T04:00:39.473479 image/svg+xml Matplotlib v3.10.8, https://matplotlib.org/

Figure 1. The five-stage pipeline as a researcher × computational lane diagram. Human decisions occupy the upper lane (corpus selection, ontology design, dimensional review, calibration audit). Computational steps occupy the lower lane (LLM coding, edge enumeration, archetype propagation). Every artefact handed across the lanes is auditable — quotes, dimensional codes, edge accumulations, dominance scores.

2026-05-24T07:36:16.126514 image/svg+xml Matplotlib v3.10.8, https://matplotlib.org/

Figure 2. Propagation from case → paper → edge, illustrated for Lampón (2022). The paper's three cases (MEVP1, MEVP2, MEVP3) are dimensionally scored on orchestration and value framing. Case-level archetype assignments aggregate to a paper-level score, which then weights every edge where Lampón is among the supporting papers. The same mechanism drives the dominance scores for all 86 active edges in the workbook.