Where every number on this site comes from.

Every figure on fuel.jrand.net is sourced from operator filings, WNA, IAEA, OECD-NEA, or NEI Magazine. Every facility row carries a data-quality flag. The full audit log is published.

Top-level figures

As of 2026-Q1 · all figures on a tU contained-uranium basis (UF₆ throughput expressed as tonnes of U)

Per-facility data quality

Conversion plants on the map

Forecast methodology

The 2024-2030 projection is computed live from each facility's status, online year, ramp years, and target utilization. Defaults reflect operator forward guidance: Cameco at its stated 12,000 tU operational target, Orano Pierrelatte at its 13,500 tU 2028 target (rather than the 15,000 nameplate ceiling), Solstice ramping to 10,500 tU on the Feb 2026 commitment, Springfields at 5,000 tU first-production-2028 with a 3-year ramp.

Western reactor demand uses the WNA Fuel Report 2025 reference scenario, with a Western-share assumption of ~68% of global demand. The 2030 anchor of 60,000 tU/yr matches the WNA Reference midpoint. Historical demand 1980-2010 is interpolated from WNA back-issues and OECD-NEA Trends in NFC; the post-Fukushima trough at 2015 (42,000) reflects Japan's collapse from ~8,000 tU pre-2011 to ~2,135 tU by 2025 per the WNA Japan profile.

The audit log (data-audit.md in the project repository) documents every correction made to the dataset since launch, with source citations and confidence flags. When a number on this site disagrees with a third-party source, the audit log shows which one was used and why.

Material flows

Mine → conversion · facility → facility

The simulator's globe can render arrows showing where uranium concentrates physically move from mine to conversion plant. Toggle "Material flows" in the globe's filter row to display them. Vertical-integration flows render in green (same parent company runs both ends). Long-term-contract flows render in blue. Chain-step flows (Malvési UF₄ → Pierrelatte UF₆) render in dashed muted gray. Mines that sell to the global spot market without a single named destination render no arrow — see the "Market mines" note below.

Market mines

Olympic Dam (BHP, AU), Honeymoon (Boss Energy, AU), Lance (Peninsula, US), and Karatau/Akbastau (KZ) sell yellowcake to the global market without a single named conversion destination. They appear on the globe as mining markers but render no arrow — their tonnage is allocated across multiple Western or Chinese buyers via spot or term contracts whose specific routing is commercially confidential.

Open data gaps

Cigar Lake and McArthur River partner shares (Orano 30/37%, Idemitsu, TEPCO) likely flow to Malvési and Japanese converters but the specific share routing is not publicly mapped. CNNC's domestic Chinese mines (e.g. Yining/Xinjiang) feed Lanzhou + Hengyang but aren't currently in our mines.json dataset.

Out of scope

One facility frequently confused with UF₆ conversion is deliberately excluded:

• NFC Hyderabad (IN) — refines yellowcake to UO₂ for India's PHWR fleet; never produces UF₆.

INB Resende (BR) appears on the simulator as an enrichment plant (~0.05M SWU/yr, ramping). Brazil's 2024 INB-Internexco contract sends Caetité U₃O₈ to Russia for conversion + enrichment, with enriched UF₆ returning to Resende by Dec 2027 for fuel fabrication.

Enrichment landscape

14 commercial plants on the map · Urenco + Orano + Centrus + Rosatom + CNNC + INB

The simulator now treats enrichment as a peer to mining and conversion (the two stages on either side of UF₆). Enrichment markers render as a paper disc with a Lucide funnel glyph and a blue rim. The Enrichment filter toggles them on and off.

The 14 plants on the map are the operating commercial enrichers as of mid-2026: Urenco × 4 (Almelo NL, Capenhurst UK, Gronau DE, Eunice US), Orano Tricastin Georges Besse II (FR), Centrus Piketon (US, HALEU demo), Rosatom × 4 (Novouralsk, Zelenogorsk, Angarsk, Seversk), CNNC × 3 (Lanzhou, Hanzhong, Emeishan), and INB Resende. Combined nameplate ~61M SWU/yr against the WNA 2025 global figure of ~64M SWU/yr.

Three deliberately excluded:

• India Rattehalli RMP — naval / weapons program, no commercial LEU output.
• Argentina Pilcaniyeu — idle since 2018, 3M SWU upgrade ambition has no funded path.
• China Heping / Plant 814 — gaseous diffusion legacy site, current centrifuge LEU capacity ambiguous.

Sanctioned facilities

Permanent treatment, not a filter

Russian conversion plants (Seversk SCC) render with a permanent red treatment — the same "Sanctioned" peer marker shown in the legend's UF₆ Conversion row. The post-2022 bloc split is the load-bearing fact of the conversion atlas. A toggle that hides it would misread the map. The flow arrows from sanctioned plants stay visible too; readers can see where Russian UF₆ actually goes.

Multi-stage facilities

Primary disc + secondary badge

A few sites do more than one stage of the cycle. CNNC Lanzhou is a captive UF₆ converter and a centrifuge enrichment plant at the same address. Seversk SCC runs UF₆ conversion and enrichment cascades on the same site. The simulator renders these as a single primary disc (the conversion stage, since this is a conversion atlas) with a small secondary badge in the upper-right notch carrying the Lucide glyph for the second stage. Toggling the Enrichment filter hides the badge along with the standalone enrichment markers.

Co-location is declared explicitly in downstream_nodes.json via a co_located_with field pointing at the conversion facility's ID. This is editorial — Tricastin's GB2 enrichment plant sits ~700 m from the Pierrelatte UF₆ line but is rendered as a separate marker because that is how Orano discloses them.

Confidence levels on flows

High · medium · low

Each flow carries a confidence label. High means a single named operator filing or regulatory document confirms the source-to-destination link. Medium means the relationship is documented at the bloc level (Cameco names "Western enrichers" generically; Urenco does not name converters in its public reporting) but not at the per-cylinder level. Low means a plausible inference based on geography and bloc, with no operator filing.

By default the simulator hides medium- and low-confidence flows so the headline pattern reads cleanly. The + less confident sub-toggle in the filter row reveals them as dashed, lower-opacity lines. The seven Cameco/Orano → Urenco supply flows fall in this bucket because no public contract disclosure pins down per-plant share. Same for Hengyang's specific destinations across the CNNC enrichment fleet.

Source list

• WNA Reactor & Uranium Requirements (Apr 2026) — global reactor count, demand
• WNA Nuclear Fuel Report 2025 — reference / lower / upper scenarios
• WNA Russia profile — Seversk capacity
• WNA China profile — Lanzhou + Jiuquan
• Cameco 2025 Annual Report
• Orano 2024 RAA
• Solstice press release, Feb 2026
• IAEA PRIS — reactor counts by country
• NEI Magazine — historical conversion review
• OECD-NEA Trends in NFC
• Public Law 118-62 (Russian Uranium Imports Act, 2024)

How to read the data-quality flags

High — sourced from the operator's own filing or a regulator-published number, vintage within 18 months. Defensible.

Medium — published in a trade source or industry report, or inferred from documented operator guidance with a known assumption. Cite-and-explain.

Low — last public disclosure is older than 5 years, or the figure is an estimate from a secondary source without operator confirmation. Use with caution and a footnote.

None — capacity is not publicly disclosed. The site shows null and explains why.