If you've ever tried to build "does this part fit my car" search for an e-commerce site, you've probably hit TecDoc — the European standard database for automotive parts fitment — without anyone actually explaining how it's structured. Here's the developer-level version. ## The core problem TecDoc solves "Does part X fit a 2018 Toyota Corolla?" is not a well-formed question. A 2018 Corolla shipped with at least four engine variants, each with different mounting points, dimensions, and part numbers for the same physical component. Fitment data has to be precise at the _engine variant_ level, not just make/model/year, or you ship wrong parts and eat return costs. TecDoc — maintained by TecAlliance, an industry consortium of OEMs and aftermarket suppliers — is the reference database that solves this at scale: 500+ vehicle makes, 900+ aftermarket brands, 400M+ fitment records. ## The data model Two linked structures: **Vehicle catalog** — a 4-level hierarchy: `Make → Model → Type → Engine` Each level gets its own TecDoc ID. The "Type" level is where it gets interesting — a 2018 Corolla 1.8 Hybrid and a 2018 Corolla 1.6 Petrol are different Type IDs, because they're different vehicles from a parts perspective even though they're the "same car" from a marketing one. **Parts catalog** — organized by _Generic Article Group_ (GAG), a standardized category code. "Brake disc" has one GAG code regardless of which of the 900+ brands submitted it. This is what makes cross-brand comparison reliable — you're never accidentally comparing a brake disc to a brake pad because two brands categorized things slightly differently. **Fitment records** are the join table: `(GAG, Type ID) → compatible`. This is the actual query your fitment search needs to run. ## Cross-reference: the other half Separately, TecDoc stores cross-reference records: `IAM part number → OEM part number`. When an aftermarket brand submits a part, they attach the OEM number it's designed to replace. This gives you multi-directional lookup — OEM number in, all equivalents out; any equivalent in, OEM number + all siblings out. The gotcha: this only works if the brand actually submitted the cross-reference. A physically identical part from a brand that skipped that step won't show up in the lookup. Production systems typically need a supplementary fitment-based matching layer to catch these gaps, and often blend in manufacturer feeds and other data sources alongside TecDoc to close coverage holes. ## What this means if you're building on top of it * Don't key your fitment queries on make/model/year alone — resolve down to Type ID (engine variant) or your compatibility data will be wrong for split-engine model years. * Quarterly update cycle means TecDoc data is never more than ~3 months stale by design. Fine for structural parts, not fine for fast-moving categories (filters, plugs) without your own sync discipline. * Direct TecDoc licensing is a real commercial process (volume commitments, TecAlliance integration) — most teams building product pages work with a data provider instead, one that combines TecDoc with other sources and ships pre-processed, query-ready output. I wrote a longer breakdown of the full data model (including how AI search engines weight fitment specificity when citing product pages) here: TecDoc Explained — I work on PartWiz, where we deal with catalog data enrichment for spare parts sellers; happy to compare notes if you're deep in this problem too. Curious what other devs have run into building fitment search — anyone dealing with ACES/PIES on the US side instead of TecDoc, would be interested to compare notes.

To predict the criticality and other reactor characteristics it is necessary to evaluate the fine-structure flux distribution caused by the heterogeneous...