Asset uptime, built from the atom — the failure-mode — up.
Every organisation in the RAM 2026 Rail SIG pre-conference feedback surfaced the same shortage — no shared failure-mode corpus, no provenance discipline, no linkage from functions to failure modes, no way to scale reliability-centered maintenance without running per-asset workshops that no operator can afford at fleet scale. This is not a tooling problem. Master-data platforms do master data well. APM platforms do asset performance well. What is missing is the content substrate underneath.
The Open Rail Corpus v1 is a public, provenance-traced reference body donated to the Rail SIG. Content is licensed CC BY 4.0. The v1 release ships a single populated scenario and two structural placeholders for v1.1.
A publicly-standardised rail freight container wagon (Sggmrs(s), 6-axle articulated bogie, UIC interoperability-compliant). Technical specifications follow UIC Leaflet 505-1/503 for loading gauges and ERA IRS 50592 for container and swap-body transport; operations under the General Contract of Use for Wagons (GCU) framework. The Swiss Federal Office of Transport Fehlerkatalog is applied — 297 failure modes across 8 chapters, each citable to the regulatory section it derives from. The regulation is published by the FOT Safety Supervision Section, Bern; authored by the Section and signed by the Head of Section.
Bundle at /scenario-a-sggmrs-container/.
A linear-asset demonstration — chainage, segments, failure-mode scoping to specific stretches of track. Public EU regulation, public corridor geometry. Structural placeholder in v1; populated in v1.1.
Governance as a feature. Scenario A already demonstrates the provenance
tiers in practice (corpus_source=bav_fk_v12_2021,
source=regulation). Scenario C formalises the vocabulary as a separate
reference artefact in v1.1.
Pull the full bundle:
curl https://openrxm.org/corpus/transport/rail/v1/scenario-a-sggmrs-container/bundle.jsonFilter via the API — Chapter 3 (Brake) failure modes at Critical risk:
curl "https://openrxm.org/corpus/transport/rail/v1/api.php?scenario=a&chapter=3&risk=Critical"Full API documentation at /api.html.
The AO Foundation (surgeryreference.aofoundation.org) solved an analogous problem in orthopedic trauma — the diffuse mass of "complex surgical cases" became a structured, citable, community-maintained reference that every surgeon, every institution, every medical-device vendor could build on. The surgery moved forward together because the substrate was published. Rail-asset reliability has never had an AO Foundation equivalent. The Open Rail Corpus is a first step toward one.
The corpus raises a question it does not yet answer: how to preserve a stable failure-mode identity across regulatory-catalog editions. When Switzerland's BAV moved the Fehlerkatalog from Version 12.0 (2021) to Version 18.0 (2026), most underlying failure mechanisms remained unchanged, but some section references shifted. A failure-mode identifier that hashes on the regulatory section breaks unnecessarily; one that hashes on the mechanism itself is more durable but harder to derive. This is the v1.1 question. Posed as a question is how good corpora start.
Join the Rail SIG working group. Use the v1 bundle in your own APM tenants. Tell us what you want to see in v2. New scenarios, new regulatory catalogs, new linear corridors, operator-anonymised observations from real pilots contribute through the SIG, evaluated for scope and provenance, integrated into subsequent releases.
I'm running. You're invited.