# AIJ UWE — cleaned data API > Wind-tunnel and field reference data from the Architectural Institute of Japan Urban Wind Environment (UWE) > benchmark, Cases A-M, cleaned into one schema, plus Eddy3D's validation against it. Static files, plain GET. Base URL: https://aij.eddy3d.com/api Fallback (raw files from the repository): https://raw.githubusercontent.com/Eddy3D-Dev/AIJ-Data-Validation/main/docs/api Repository: https://github.com/Eddy3D-Dev/AIJ-Data-Validation Release: 3.11.3 (2026-11-09). This is the latest alias, which always equals the newest release; the frozen copy of 3.11.3 is https://aij.eddy3d.com/api/3.11.3. All releases: https://aij.eddy3d.com/api/versions.json API schema version: 2.6.0. Data version (SHA-256 of the served data): 7ee0afb7cec8197e7edde0e7c282c8e6e9670b7cef003e6aa218c5a536817f48. Licence: CC-BY-4.0. Data: Kikumoto et al. (2026); the collectors, DOI and requested citation of each case are in cases/{id}.json. ## How to use this 1. GET index.json — every endpoint and file (with size and SHA-256), versions, changelog, citation. 2. GET cases.json — every case with its frame, units, counts and links. Pick an id (A, B, C, D, E, F, G, H, I, J, K, L, M). 3. GET cases/{id}.json — read `frame`, `units` and `known_issues` BEFORE the numbers; they differ per case. 4. GET cases/{id}/points.json (or points.csv) — every value is SI (m, m/s, m^2/s^2; concentrations in ppm, temperatures in degC) in ONE frame per case; unmeasured values are null. Cases E, F and G publish dimensionless ratios (velocity_ratio, k_ratio), not speeds in m/s; Case M has wind-tunnel and field rows (`campaign`). Read `units` and the normalisation first. 5. GET cases/{id}/geometry.stl — ASCII STL in metres, outward normals, in the case's own frame (cases/{id}.json `frame`). Cases A-D and H-L are model-scale with +x along the wind at WD = 0; E to G and M are full-scale: E and F keep their geographic frame (+x east, +y north), G has +x downwind, and M has +x north and +y west. geometry.json lists the boxes where the geometry IS boxes (A, B, C, D, G, H, I, J, K, L), so you need not parse the STL at all. 6. GET validation.json for every case and engine at once, or cases/{id}/validation.json for one case in full. Quote the `headline` scores. An engine whose status is not "scored" has NO result for that case: do not infer one. Each result's `pairs` is the per-point measured and modelled values, so any score can be recomputed. Status: A: openfoam12-realizableKE scored, openlb scored, fluidx3d scored, ml-surrogate scored; B: openfoam12-realizableKE scored, openlb scored, fluidx3d scored, ml-surrogate scored; C: openfoam12-realizableKE scored, openlb not comparable, fluidx3d pending, ml-surrogate scored; D: openfoam12-realizableKE scored, openlb not comparable, fluidx3d pending, ml-surrogate scored; E: openfoam12-realizableKE scored, openlb not comparable, fluidx3d pending, ml-surrogate scored; F: openfoam12-realizableKE scored, openlb pending, fluidx3d pending, ml-surrogate pending; G: openfoam12-realizableKE scored, openlb pending, fluidx3d pending, ml-surrogate pending; H: openfoam12-realizableKE scored, openlb pending, fluidx3d pending, ml-surrogate pending; I: openfoam12-realizableKE scored, openfoam12-kEpsilon scored, openfoam12-kOmegaSST scored, openlb pending, fluidx3d pending, ml-surrogate pending; J: openfoam12-realizableKE pending, umf-realizableKE scored, umf-buoyantKEpsilon scored, openlb pending, fluidx3d pending, ml-surrogate pending; K: openfoam12-realizableKE scored, openlb pending, fluidx3d pending, ml-surrogate pending; L: openfoam12-realizableKE pending, umf-realizableKE scored, openlb pending, fluidx3d pending, ml-surrogate pending; M: openfoam12-realizableKE scored, openlb pending, fluidx3d pending, ml-surrogate pending. 7. All geometry at once: GET geometry.json (every STL with its case, configuration, units, frame and SHA-256) or geometry.zip. Each STL stays in its own case's frame and units; they are NOT one scene. 8. Every JSON document names its JSON Schema in `$schema` (schemas/{name}.schema.json); openapi.json describes every endpoint. Download URLs carry ?v=: use the URLs the documents give, not hand-built ones. 9. Send a descriptive User-Agent: the site's CDN answers HTTP 403 to Python urllib's default one and to some AI crawlers. If you are refused, the fallback URL above serves byte-identical files from the repository. 10. To make a result reproducible, pin the release: GET versions.json, take the base_url of the release you used, and fetch every path below from it instead of from the unversioned base. A release's files never change, and every URL to an API file inside its documents points back into it. Quote the release number with any number you report. 11. To score YOUR OWN solver or model: GET cases/{id}/probes.csv (the points to probe and what each is scored as), probe your field there, add your values as columns (U V W speed k cstar theta T), and score with validation/score_submissions.py from a checkout of the repository (uv run python validation/score_submissions.py ; nothing beyond pyproject.toml), which applies cases/{id}.json `scoring` (reference speed, conventions) with the metrics that score the served results. cases/{id}.json `inflow` is the inlet contract: `benchmark_data` is the measured profile, `recommended` is this repository's choice (a log law from the fit, k and epsilon from the ABL relations); cases/{id}/inflow_openfoam.txt is it as OpenFOAM files. The walk is https://aij.eddy3d.com/byos/. 12. Community results: GET submissions.json (every submitted solver, scored by this repository's build from the values it gave at the published points; self-reported inputs, scored here) and submissions/{slug}.json. They are not Eddy3D's results: those stay in validation.json. A submission marked example is the repository's own worked example. ## Endpoints - https://aij.eddy3d.com/api/index.json — every endpoint and file with its size and SHA-256, versions, compatibility policy, changelog, licence, how to cite the API (cite.api) - https://aij.eddy3d.com/api/versions.json — every release, newest first: version, date, status (latest or superseded), schema and data version, base URL, tag, changelog; always at the unversioned /api/ path - https://aij.eddy3d.com/api/cases.json — every case: id, title, frame, units, counts, engine status, links - cases/{id}.json (base URL + path, one file per value) — one case in full: configurations, frame, units, inflow fit and approach-flow profile, point-table summary, geometry files, validation status, known data issues, attribution, citation of the case's dataset: - https://aij.eddy3d.com/api/cases/A.json - https://aij.eddy3d.com/api/cases/B.json - https://aij.eddy3d.com/api/cases/C.json - https://aij.eddy3d.com/api/cases/D.json - https://aij.eddy3d.com/api/cases/E.json - https://aij.eddy3d.com/api/cases/F.json - https://aij.eddy3d.com/api/cases/G.json - https://aij.eddy3d.com/api/cases/H.json - https://aij.eddy3d.com/api/cases/I.json - https://aij.eddy3d.com/api/cases/J.json - https://aij.eddy3d.com/api/cases/K.json - https://aij.eddy3d.com/api/cases/L.json - https://aij.eddy3d.com/api/cases/M.json - cases/{id}/points.json (base URL + path, one file per value) — the cleaned measurement table: SI units, one frame, null where not measured: - https://aij.eddy3d.com/api/cases/A/points.json - https://aij.eddy3d.com/api/cases/B/points.json - https://aij.eddy3d.com/api/cases/C/points.json - https://aij.eddy3d.com/api/cases/D/points.json - https://aij.eddy3d.com/api/cases/E/points.json - https://aij.eddy3d.com/api/cases/F/points.json - https://aij.eddy3d.com/api/cases/G/points.json - https://aij.eddy3d.com/api/cases/H/points.json - https://aij.eddy3d.com/api/cases/I/points.json - https://aij.eddy3d.com/api/cases/J/points.json - https://aij.eddy3d.com/api/cases/K/points.json - https://aij.eddy3d.com/api/cases/L/points.json - https://aij.eddy3d.com/api/cases/M/points.json - cases/{id}/points.csv (base URL + path, one file per value) — the same table as CSV (empty cell = not measured): - https://aij.eddy3d.com/api/cases/A/points.csv?v=a46201996b22 - https://aij.eddy3d.com/api/cases/B/points.csv?v=dec97c21217a - https://aij.eddy3d.com/api/cases/C/points.csv?v=7448f0ceae37 - https://aij.eddy3d.com/api/cases/D/points.csv?v=596246341e73 - https://aij.eddy3d.com/api/cases/E/points.csv?v=e6e89432e706 - https://aij.eddy3d.com/api/cases/F/points.csv?v=09b109e58dd3 - https://aij.eddy3d.com/api/cases/G/points.csv?v=914883bf7fec - https://aij.eddy3d.com/api/cases/H/points.csv?v=ef5fafc61b8e - https://aij.eddy3d.com/api/cases/I/points.csv?v=2864fee9eb19 - https://aij.eddy3d.com/api/cases/J/points.csv?v=444b7aaa1791 - https://aij.eddy3d.com/api/cases/K/points.csv?v=8708f3251861 - https://aij.eddy3d.com/api/cases/L/points.csv?v=b575fe11d107 - https://aij.eddy3d.com/api/cases/M/points.csv?v=8d32dc8f762b - cases/{id}/approach_flow.csv (base URL + path, one file per value) — the measured approach-flow profile: z (m), U (m/s), and k (m^2/s^2) or u_rms (m/s) where published; Case G's U and k are dimensionless (U/Ub, k/Ub^2): - https://aij.eddy3d.com/api/cases/A/approach_flow.csv?v=2f3b37fa8e43 - https://aij.eddy3d.com/api/cases/B/approach_flow.csv?v=126a620f5698 - https://aij.eddy3d.com/api/cases/C/approach_flow.csv?v=98849731e877 - https://aij.eddy3d.com/api/cases/D/approach_flow.csv?v=c8239428b9b9 - https://aij.eddy3d.com/api/cases/E/approach_flow.csv?v=4971944cd8e9 - https://aij.eddy3d.com/api/cases/F/approach_flow.csv?v=66aefe558a82 - https://aij.eddy3d.com/api/cases/G/approach_flow.csv?v=45a11857cdf9 - https://aij.eddy3d.com/api/cases/H/approach_flow.csv?v=ce516d4054a5 - https://aij.eddy3d.com/api/cases/I/approach_flow.csv?v=8c9a2fa1669a - https://aij.eddy3d.com/api/cases/J/approach_flow.csv?v=076668c06f1c - https://aij.eddy3d.com/api/cases/K/approach_flow.csv?v=5d197b5001ba - https://aij.eddy3d.com/api/cases/L/approach_flow.csv?v=c428993827d4 - https://aij.eddy3d.com/api/cases/M/approach_flow.csv?v=0042caa39493 - cases/{id}/geometry.json (base URL + path, one file per value) — the case's STL files with statistics, renders, and the boxes where the geometry is boxes: - https://aij.eddy3d.com/api/cases/A/geometry.json - https://aij.eddy3d.com/api/cases/B/geometry.json - https://aij.eddy3d.com/api/cases/C/geometry.json - https://aij.eddy3d.com/api/cases/D/geometry.json - https://aij.eddy3d.com/api/cases/E/geometry.json - https://aij.eddy3d.com/api/cases/F/geometry.json - https://aij.eddy3d.com/api/cases/G/geometry.json - https://aij.eddy3d.com/api/cases/H/geometry.json - https://aij.eddy3d.com/api/cases/I/geometry.json - https://aij.eddy3d.com/api/cases/J/geometry.json - https://aij.eddy3d.com/api/cases/K/geometry.json - https://aij.eddy3d.com/api/cases/L/geometry.json - https://aij.eddy3d.com/api/cases/M/geometry.json - cases/{id}/geometry.stl (base URL + path, one file per value) — cleaned ASCII STL of the default configuration: metres, case frame, outward normals: - https://aij.eddy3d.com/api/cases/A/geometry.stl?v=89acdfbbff84 - https://aij.eddy3d.com/api/cases/B/geometry.stl?v=e6de70eeb17d - https://aij.eddy3d.com/api/cases/C/geometry.stl?v=b8c31c57c453 - https://aij.eddy3d.com/api/cases/D/geometry.stl?v=531895531271 - https://aij.eddy3d.com/api/cases/E/geometry.stl?v=c29624fdae6d - https://aij.eddy3d.com/api/cases/F/geometry.stl?v=8799740aa324 - https://aij.eddy3d.com/api/cases/G/geometry.stl?v=09d23c794155 - https://aij.eddy3d.com/api/cases/H/geometry.stl?v=4836687efff9 - https://aij.eddy3d.com/api/cases/I/geometry.stl?v=ff61cbb22798 - https://aij.eddy3d.com/api/cases/J/geometry.stl?v=1bd33d8d3531 - https://aij.eddy3d.com/api/cases/K/geometry.stl?v=f0f8a3adfa29 - https://aij.eddy3d.com/api/cases/L/geometry.stl?v=20864bc75187 - https://aij.eddy3d.com/api/cases/M/geometry.stl?v=e4e0a2044536 - cases/{id}/geometry_{configuration}.stl (base URL + path, one file per value) — the other configurations' STLs (C: 0D, 1D; E: bc, before construction): - https://aij.eddy3d.com/api/cases/C/geometry_0D.stl?v=0f08bcd4dd6b - https://aij.eddy3d.com/api/cases/C/geometry_1D.stl?v=ac92c8ed427c - https://aij.eddy3d.com/api/cases/E/geometry_bc.stl?v=9b944d97a817 - cases/{id}/probes.csv (base URL + path, one file per value) — the published points to probe for a comparison: the key columns of a scored file, x y z in the case frame, and what each row is scored as: - https://aij.eddy3d.com/api/cases/A/probes.csv?v=b2a57394ad1d - https://aij.eddy3d.com/api/cases/B/probes.csv?v=025b8113dcde - https://aij.eddy3d.com/api/cases/C/probes.csv?v=5b794e8356f0 - https://aij.eddy3d.com/api/cases/D/probes.csv?v=4617e9e0d4e1 - https://aij.eddy3d.com/api/cases/E/probes.csv?v=e248607ca6fe - https://aij.eddy3d.com/api/cases/F/probes.csv?v=d7dbd73343b9 - https://aij.eddy3d.com/api/cases/G/probes.csv?v=219c5a86d3e8 - https://aij.eddy3d.com/api/cases/H/probes.csv?v=450e304dc11d - https://aij.eddy3d.com/api/cases/I/probes.csv?v=67f76ec18f93 - https://aij.eddy3d.com/api/cases/J/probes.csv?v=da4854c6791c - https://aij.eddy3d.com/api/cases/K/probes.csv?v=16f8836953ce - https://aij.eddy3d.com/api/cases/L/probes.csv?v=abee3cd8c088 - https://aij.eddy3d.com/api/cases/M/probes.csv?v=877ae498c348 - cases/{id}/inflow_openfoam.txt (base URL + path, one file per value) — the recommended inlet as OpenFOAM 12 files in one text (0/U, 0/k, 0/epsilon, 0/nut, system/probes with the case's points); the benchmark's data is the measured profile, the inlet is this repository's choice: - https://aij.eddy3d.com/api/cases/A/inflow_openfoam.txt?v=08b4a3bd4448 - https://aij.eddy3d.com/api/cases/B/inflow_openfoam.txt?v=c32ae01f7d37 - https://aij.eddy3d.com/api/cases/C/inflow_openfoam.txt?v=c11db5477c30 - https://aij.eddy3d.com/api/cases/D/inflow_openfoam.txt?v=032de3c55e72 - https://aij.eddy3d.com/api/cases/E/inflow_openfoam.txt?v=bf8f2db5c3b1 - https://aij.eddy3d.com/api/cases/F/inflow_openfoam.txt?v=8a2b2c54fcbb - https://aij.eddy3d.com/api/cases/G/inflow_openfoam.txt?v=71c290c57b9e - https://aij.eddy3d.com/api/cases/H/inflow_openfoam.txt?v=444636fc0f47 - https://aij.eddy3d.com/api/cases/I/inflow_openfoam.txt?v=93c648d3805b - https://aij.eddy3d.com/api/cases/J/inflow_openfoam.txt?v=4ac58de3bf92 - https://aij.eddy3d.com/api/cases/K/inflow_openfoam.txt?v=5292186a727d - https://aij.eddy3d.com/api/cases/L/inflow_openfoam.txt?v=894ba765b86f - https://aij.eddy3d.com/api/cases/M/inflow_openfoam.txt?v=2f707ce4d6f1 - cases/{id}/validation.json (base URL + path, one file per value) — Eddy3D's scores per engine and run, with figures and the per-point pairs: - https://aij.eddy3d.com/api/cases/A/validation.json - https://aij.eddy3d.com/api/cases/B/validation.json - https://aij.eddy3d.com/api/cases/C/validation.json - https://aij.eddy3d.com/api/cases/D/validation.json - https://aij.eddy3d.com/api/cases/E/validation.json - https://aij.eddy3d.com/api/cases/F/validation.json - https://aij.eddy3d.com/api/cases/G/validation.json - https://aij.eddy3d.com/api/cases/H/validation.json - https://aij.eddy3d.com/api/cases/I/validation.json - https://aij.eddy3d.com/api/cases/J/validation.json - https://aij.eddy3d.com/api/cases/K/validation.json - https://aij.eddy3d.com/api/cases/L/validation.json - https://aij.eddy3d.com/api/cases/M/validation.json - cases/{id}/pairs/{engine}.csv (base URL + path, one file per value) — per-point measured and modelled values of one engine's runs on the case (columns described by the result's pairs_columns): - https://aij.eddy3d.com/api/cases/A/pairs/fluidx3d.csv?v=fb16be8c5800 - https://aij.eddy3d.com/api/cases/A/pairs/ml-surrogate.csv?v=6177f32a94f4 - https://aij.eddy3d.com/api/cases/A/pairs/openfoam12-realizableKE.csv?v=0695411d68cf - https://aij.eddy3d.com/api/cases/A/pairs/openlb.csv?v=7376660704bd - https://aij.eddy3d.com/api/cases/B/pairs/fluidx3d.csv?v=c4af803fd09e - https://aij.eddy3d.com/api/cases/B/pairs/ml-surrogate.csv?v=eb9a9c3970e7 - https://aij.eddy3d.com/api/cases/B/pairs/openfoam12-realizableKE.csv?v=9f7c10ff5544 - https://aij.eddy3d.com/api/cases/B/pairs/openlb.csv?v=bdd88dbe936b - https://aij.eddy3d.com/api/cases/C/pairs/ml-surrogate.csv?v=6a2b60658913 - https://aij.eddy3d.com/api/cases/C/pairs/openfoam12-realizableKE.csv?v=f3d2b649b543 - https://aij.eddy3d.com/api/cases/D/pairs/ml-surrogate.csv?v=9dd71c9ce66b - https://aij.eddy3d.com/api/cases/D/pairs/openfoam12-realizableKE.csv?v=fab4f6f21ad5 - https://aij.eddy3d.com/api/cases/E/pairs/ml-surrogate.csv?v=90b71f5762aa - https://aij.eddy3d.com/api/cases/E/pairs/openfoam12-realizableKE.csv?v=fa16bc3bcd62 - https://aij.eddy3d.com/api/cases/F/pairs/openfoam12-realizableKE.csv?v=e81e010859e6 - https://aij.eddy3d.com/api/cases/G/pairs/openfoam12-realizableKE.csv?v=fda9fe2f217e - https://aij.eddy3d.com/api/cases/H/pairs/openfoam12-realizableKE.csv?v=915d369f503f - https://aij.eddy3d.com/api/cases/I/pairs/openfoam12-kEpsilon.csv?v=6e6ba0f446e8 - https://aij.eddy3d.com/api/cases/I/pairs/openfoam12-kOmegaSST.csv?v=549a7aa4117e - https://aij.eddy3d.com/api/cases/I/pairs/openfoam12-realizableKE.csv?v=d1bafd9c68e2 - https://aij.eddy3d.com/api/cases/J/pairs/umf-buoyantKEpsilon.csv?v=87ba413bfa96 - https://aij.eddy3d.com/api/cases/J/pairs/umf-realizableKE.csv?v=cb34fb5fc9c7 - https://aij.eddy3d.com/api/cases/K/pairs/openfoam12-realizableKE.csv?v=746a236c4635 - https://aij.eddy3d.com/api/cases/L/pairs/umf-realizableKE.csv?v=a0aedc557c46 - https://aij.eddy3d.com/api/cases/M/pairs/openfoam12-realizableKE.csv?v=9ea7d94ddd27 - https://aij.eddy3d.com/api/validation.json — every case and engine in one table: status, runs, headline scores - https://aij.eddy3d.com/api/geometry.json — every STL of every case in one list: case, configuration, units, frame, bounding box, size, SHA-256 - https://aij.eddy3d.com/api/geometry.zip — every STL in one archive (each keeps its own case's frame and units: they are not one scene) - https://aij.eddy3d.com/api/submissions.json — the community results: every submitted solver with its headline scores per case, scored by this build from the submitted values - submissions/{slug}.json (base URL + path, one file per value) — one submission in full: its submission.yaml, and every score, coverage and group per case: - https://aij.eddy3d.com/api/submissions/eddy3d-of12-casee-refined-mesh.json - https://aij.eddy3d.com/api/submissions/eddy3d-of12-casek-workbook-inflow.json - https://aij.eddy3d.com/api/submissions/eddy3d-of12-measured-inlet.json - https://aij.eddy3d.com/api/submissions/eddy3d-of12-mesh2x-measured-inlet.json - https://aij.eddy3d.com/api/submissions/eddy3d-of12-nutcap-lifted.json - https://aij.eddy3d.com/api/submissions/eddy3d-of12-sst-urans.json - https://aij.eddy3d.com/api/submissions/example-openfoam12.json - submissions/{slug}/{id}.csv (base URL + path, one file per value) — the values a submission gave at the published points of one case (the key columns of probes.csv plus its value columns): - https://aij.eddy3d.com/api/submissions/eddy3d-of12-casee-refined-mesh/E.csv?v=c302ea867647 - https://aij.eddy3d.com/api/submissions/eddy3d-of12-casek-workbook-inflow/K.csv?v=fd83799198fb - https://aij.eddy3d.com/api/submissions/eddy3d-of12-measured-inlet/A.csv?v=cdc0dad39c20 - https://aij.eddy3d.com/api/submissions/eddy3d-of12-measured-inlet/B.csv?v=0f5b2d79329b - https://aij.eddy3d.com/api/submissions/eddy3d-of12-measured-inlet/H.csv?v=222411bc15a1 - https://aij.eddy3d.com/api/submissions/eddy3d-of12-mesh2x-measured-inlet/A.csv?v=d21d791d27b1 - https://aij.eddy3d.com/api/submissions/eddy3d-of12-nutcap-lifted/A.csv?v=d55ee1f8ca3c - https://aij.eddy3d.com/api/submissions/eddy3d-of12-nutcap-lifted/B.csv?v=d35fb0a6cd1b - https://aij.eddy3d.com/api/submissions/eddy3d-of12-nutcap-lifted/C.csv?v=35838affb859 - https://aij.eddy3d.com/api/submissions/eddy3d-of12-nutcap-lifted/D.csv?v=2dfde16debe1 - https://aij.eddy3d.com/api/submissions/eddy3d-of12-nutcap-lifted/E.csv?v=7adb9c8af800 - https://aij.eddy3d.com/api/submissions/eddy3d-of12-nutcap-lifted/F.csv?v=529a40fd2c5b - https://aij.eddy3d.com/api/submissions/eddy3d-of12-nutcap-lifted/G.csv?v=bc0ad9ed02d5 - https://aij.eddy3d.com/api/submissions/eddy3d-of12-nutcap-lifted/H.csv?v=976a8b51df69 - https://aij.eddy3d.com/api/submissions/eddy3d-of12-nutcap-lifted/K.csv?v=e26fa1b76a4e - https://aij.eddy3d.com/api/submissions/eddy3d-of12-nutcap-lifted/L.csv?v=56a0918b8c4b - https://aij.eddy3d.com/api/submissions/eddy3d-of12-nutcap-lifted/M.csv?v=6fd56831269c - https://aij.eddy3d.com/api/submissions/eddy3d-of12-sst-urans/A.csv?v=98a4d275b3f2 - https://aij.eddy3d.com/api/submissions/example-openfoam12/A.csv?v=4eec5b1fe73a - schemas/{name}.schema.json (base URL + path, one file per value) — the JSON Schema (draft 2020-12) of each JSON document: - https://aij.eddy3d.com/api/schemas/case.schema.json - https://aij.eddy3d.com/api/schemas/cases.schema.json - https://aij.eddy3d.com/api/schemas/geometry-index.schema.json - https://aij.eddy3d.com/api/schemas/geometry.schema.json - https://aij.eddy3d.com/api/schemas/index.schema.json - https://aij.eddy3d.com/api/schemas/points.schema.json - https://aij.eddy3d.com/api/schemas/submission.schema.json - https://aij.eddy3d.com/api/schemas/submissions.schema.json - https://aij.eddy3d.com/api/schemas/validation-summary.schema.json - https://aij.eddy3d.com/api/schemas/validation.schema.json - https://aij.eddy3d.com/api/schemas/versions.schema.json - https://aij.eddy3d.com/api/openapi.json — OpenAPI 3.1 description of every endpoint - https://aij.eddy3d.com/api/llms.txt — plain-text entry point for agents ## Cases - A: Case A — 1:1:2 building. Isolated 1:1:2 building, velocity field (split-fibre hot wire). 186 rows, 186 points, columns point_id, config, instrument, wind_direction_deg, x, y, z, U, V, W, uu, vv, ww, k, campaign. Engines: openfoam12-realizableKE scored, openlb scored, fluidx3d scored, ml-surrogate scored. - B: Case B — 1:4:4 slab. Isolated 1:4:4 slab, velocity field (split-fibre hot wire). 224 rows, 224 points, columns point_id, config, instrument, wind_direction_deg, x, y, z, U, V, W, uu, vv, ww, k, campaign. Engines: openfoam12-realizableKE scored, openlb scored, fluidx3d scored, ml-surrogate scored. - C: Case C — 3 × 3 block array. 3 x 3 array of 0.20 m cubes, centre block at 0D / 1D / 2D, pedestrian-height scalar speed (thermistor), 3 wind directions. 1080 rows, 120 points, columns point_id, config, instrument, wind_direction_deg, x, y, z, Vs, campaign. Engines: openfoam12-realizableKE scored, openlb not comparable, fluidx3d pending, ml-surrogate scored. - D: Case D — high-rise among city blocks. 1:1:4 tower in a regular array of 82 low blocks, pedestrian-height scalar speed (thermistor, 3 wind directions) plus a 33-point split-fibre velocity field at WD = 0. 267 rows, 78 points, columns point_id, config, instrument, wind_direction_deg, x, y, z, U, V, W, uu, vv, ww, k, Vs, campaign. Engines: openfoam12-realizableKE scored, openlb not comparable, fluidx3d pending, ml-surrogate scored. - E: Case E — Niigata district. Real district in Niigata at 1:250, before and after a 60 m high-rise and two 18 m mid-rises; pedestrian-height velocity ratio (thermistor), 16 wind directions. 2560 rows, 80 points, columns point_id, config, instrument, wind_direction, wind_direction_deg, x, y, z, velocity_ratio, campaign. Engines: openfoam12-realizableKE scored, openlb not comparable, fluidx3d pending, ml-surrogate scored. - F: Case F — Shinjuku district. Shinjuku sub-centre, Tokyo, November 1978: 1:800 tunnel (thermistor, 5 / 9 / 20 m) and a year of field cup anemometers; velocity ratio to the 500 m wind, 12 tunnel and 16 field directions. 472 rows, 13 points, columns point_id, config, instrument, wind_direction, wind_direction_deg, x, y, z, velocity_ratio, campaign, velocity_ratio_sd. Engines: openfoam12-realizableKE scored, openlb pending, fluidx3d pending, ml-surrogate pending. - G: Case G — flow behind a tree belt. Field observation behind the Tsuiji-matsu tree belt (a porous windbreak, 74 m × 2 m × 7 m with a 1.2 m gap below): normalised mean speed and k at 28 points on the centre section (sonic anemometers). 28 rows, 28 points, columns point_id, config, instrument, wind_direction_deg, x, y, z, velocity_ratio, campaign, k_ratio. Engines: openfoam12-realizableKE scored, openlb pending, fluidx3d pending, ml-surrogate pending. - H: Case H — 1:1:2 building with a floor point source. Isolated 1:1:2 building with an ethylene point source on the floor behind it: velocity (split-fibre hot wire) and concentration (hydrocarbon analyser) at 876 points. 1652 rows, 876 points, columns point_id, config, instrument, wind_direction_deg, x, y, z, U, V, W, uu, vv, ww, k, campaign, C, c_rms, concentration_ratio. Engines: openfoam12-realizableKE scored, openlb pending, fluidx3d pending, ml-surrogate pending. - I: Case I — cube with a roof point source. Isolated cube with an ethane point source at the centre of its roof: velocity (laser Doppler, 91 points) and concentration (hydrocarbon analyser, 539 points). 630 rows, 630 points, columns point_id, config, instrument, wind_direction_deg, x, y, z, U, V, W, k, campaign, C, concentration_ratio. Engines: openfoam12-realizableKE scored, openfoam12-kEpsilon scored, openfoam12-kOmegaSST scored, openlb pending, fluidx3d pending, ml-surrogate pending. - J: Case J — 1:1:2 building, non-isothermal. Isolated 1:1:2 building with a floor tracer source in an unstably stratified boundary layer: velocity, temperature, tracer concentration and their turbulent fluxes at 444 points on three planes. 444 rows, 444 points, columns point_id, config, instrument, wind_direction_deg, x, y, z, U, V, W, uu, vv, ww, k, campaign, C, c_rms, concentration_ratio, T, T_rms, uT, vT, wT. Engines: openfoam12-realizableKE pending, umf-realizableKE scored, umf-buoyantKEpsilon scored, openlb pending, fluidx3d pending, ml-surrogate pending. - K: Case K — 9 x 9 cube array, point source. 9 x 9 array of 0.06 m cubes with a tracer point source on the floor between rows 2 and 3: velocity and tracer concentration at 203 points in the streets and over the first rows. 203 rows, 203 points, columns point_id, config, instrument, wind_direction_deg, x, y, z, U, V, W, uu, vv, ww, k, campaign, C, c_rms, concentration_ratio. Engines: openfoam12-realizableKE scored, openlb pending, fluidx3d pending, ml-surrogate pending. - L: Case L — 9 x 14 cube array, line source, non-isothermal. 14 rows by 9 columns of 0.06 m cubes with a tracer line source on the floor across the tunnel, in an unstably stratified boundary layer: velocity, temperature, concentration and their fluxes at 344 points in the streets and over the roofs. 344 rows, 344 points, columns point_id, config, instrument, wind_direction_deg, x, y, z, U, V, W, uu, vv, ww, k, campaign, uv, uw, C, c_rms, concentration_ratio, T, T_rms, uc, vc, wc, uT, vT, wT. Engines: openfoam12-realizableKE pending, umf-realizableKE scored, openlb pending, fluidx3d pending, ml-surrogate pending. - M: Case M — TPU Atsugi campus. Tracer-gas release on the Tokyo Polytechnic University Atsugi campus: 1:600 tunnel (velocity at 180°, ethylene concentration at 175°, 180° and 185°) and 18 field runs with PMCH, 15 points 3 m above ground. 330 rows, 15 points, columns point_id, config, instrument, wind_direction_deg, x, y, z, U, V, W, uu, vv, ww, k, campaign, run, C, c_rms, concentration_ratio. Engines: openfoam12-realizableKE scored, openlb pending, fluidx3d pending, ml-surrogate pending. ## Known data issues - [A] Lateral velocity V is positive toward +y (checked: V = -2.893 m/s at (-0.040, -0.050, 0.010), just outside the y < 0 side face, deflecting away from the centreline). Case B uses the OPPOSITE sign; the API flips Case B so both agree. - [A] Coordinates are metres (Case B's are millimetres in the published file). - [A] Case A's turbulence columns are rms values (u_rms, ...). The API serves VARIANCES (uu = u_rms^2) so A and B share one column meaning. - [B] The published coordinates are MILLIMETRES (x[mm]) while velocities are m/s. The API serves metres (coordinate_scale_applied = 0.001). Read as metres, every probe lands 1000x too far out. - [B] The published lateral velocity V is positive toward -y (V = +3.630 m/s at (-0.025, -0.110, 0.013), just outside the y < 0 side face where the flow must deflect toward -y). The API multiplies V by -1 (lateral_sign_applied = -1) so that V is positive toward +y as in Case A. Scored without this, Case B's V comes out at R = -0.910 against R = +0.959 for U on the same points. - [B] The RANS solution is symmetric, so model V = 0 on the centre plane while the tunnel measures small non-zero values there. - [C] The measured quantity Vs is the MEAN OMNIDIRECTIONAL SCALAR speed from a thermistor, not |mean velocity vector|. A RANS |U| is biased low against it; the data paper suggests comparing sqrt(|U|^2 + 2k) as a conservative correction (over all 1080 rows no Vs is below 0.13 U_H). - [C] The 'CB' column is TEXT ('0D', '1D', '2D'); a reader that parses every field as a number gets NaN for it. The header ends with a trailing comma (an empty 8th column). - [C] 1080 rows = 3 wind directions x 3 centre buildings x 120 points, not 3 configurations. In the 0D configuration the 25 centre grid nodes are still absent although the ground there is free: the grid is held constant across configurations. - [C] 24 of the 120 points sit exactly on a building face plane (|x| = 0.30 or |y| = 0.30) in the open gap between two blocks: legitimate fluid points, but a probe's cell lookup is delicate there. - [C] Point numbers 1, 2, 3, 34, 35, 36, 37, 46, 55, 100, 101, 102 do not occur (No. runs 4 ... 132 with nine interior gaps). - [C] Experimental repeatability, from the mirror-symmetric point pairs: mean |dVs| 0.088-0.137 m/s (0.024-0.038 U_H). - [D] All of Case D's data lives in LF_caseD.xls, an OLE2/BIFF8 workbook (read it with xlrd 2.x; openpyxl cannot). Sheets: 'Geometry&Points', 'Inflow', 'Results (Exp_S)' (split fibre, WD = 0), 'Results (Exp_T)' (thermistor, 0 / 22.5 / 45 deg). - [D] Coordinates are millimetres in the workbook; the API serves metres. - [D] The shipped BD_caseD.stl has 1572 facets of which 576 are duplicates, and all 83 roofs are wound inward (signed volume 0.011159 m^3 against the true 0.021477 m^3). The API's geometry.stl is rebuilt from the 83 shells' bounding boxes: 996 facets, watertight, volume 0.021477 m^3. - [D] The thermistor reads a SCALAR speed. Over the 33 points where both instruments overlap at WD = 0, the ratio of mean thermistor speed to mean |U| is 1.342 (the mean of the per-point ratios is 1.50). sqrt(|U|^2 + 2k) from the split-fibre data matches the thermistor to FB +0.058, R 0.947; |U| alone to FB +0.292, R 0.872. - [D] Unmeasured thermistor cells are '-' in the workbook (null here). At WD = 0 the 66 measured values are 33 exact mirror pairs, so only 33 are independent; at 22.5 and 45 deg all 66 are independent. - [D] A prior Eddy3D comparison kept in the folder (case-d-eddy3d-prior-comparison.xlsx) scores R = +0.096 against the thermistor. The score comes from a frame or geometry error; do not use it as a validation. - [E] The velocity ratio is normalised by the inflow velocity at 15.9 m (= 3.935 m/s), NOT by U_H and NOT by UR = 7.8 m/s. A prior Eddy3D comparison in the folder divided by the wrong reference and came out low by a factor of 2.048 (FAC2 0.338 -> 0.863 once rescaled, R unchanged at 0.661). - [E] All 2560 published ratios are integer multiples of 0.0198204 (55 distinct levels): the data's own resolution is +/-0.0099, 3.1 % of the median value and about 10 % at the lowest points. - [E] The readme describes AF_caseE.csv, RS_caseE.csv and BD_caseE.stl, but this repository holds only the .3dm, .ghx, .xls and the MP drawing; everything here is extracted from the workbook and the 3dm. - [E] case-e.3dm is FULL-SCALE metres (MP_caseE.png is drawn at model scale, +/-0.8 m). Use layers 'buildings' (5,015 meshes, before construction) and 'buildings_new_construction' (3 prisms); exclude the map plate on the Japanese-named general layer, the terrain plates (the site is flat) and 'probes'. - [E] case-e.3dm: 1,563 of 8,436 triangles (18.5 %) of the building meshes are wound against their neighbours; all are re-wound here. - [E] Do not take the measurement coordinates from the 3dm's 'probes' layer: its 80 spheres sit a median 2.31 m (max 21 m) from the workbook's points, which are authoritative (all 80 fall in open space, min 0.83 m from a building edge). - [E] The measured quantity is a scalar (thermistor) speed; a RANS |U| is biased low against it in a dense canopy. The construction signal (ac - bc) is largest for SW (mean |d| 0.2155) and smallest for SSE (0.093). - [F] The 16 field rows of point 14 in RS-FO_caseF.csv hold the workbook's value of the NEXT wind direction clockwise (the series is shifted by one); the cleaned table follows the workbook (clean/QUESTIONS.md QF1). - [F] Tunnel and STL coordinates are MODEL scale (1:800); the field's are full scale. The case frame is full scale; the tunnel's x, y, z and the STL are multiplied by 800. - [F] z is height above the tunnel floor. The readme's rule 'ground = z - 10 m' gives the 5 / 9 / 20 m planes, but the STL's terrain under all 13 points is at z = 5 m, so the planes stand 10 / 14 / 25 m above it. - [F] BD_caseF.stl is wound inward on every ground, roof and sampled wall facet and is open along the rim of its 725 m disc (812 boundary edges); the served geometry.stl is re-wound outward and closed with a base slab down to z = -1 m. - [F] MP_caseF.csv names the two reference points the other way round from the workbook, the STL and MP_caseF.png (the workbook's Shinjuku-Mitsui Building is the 225 m tower at (-17.5, 96.0), the KDD Building the 180 m one at (85.8, -347.1)). - [F] Field values are the legacy workbook's roof-anemometer ratios times 0.940 (NW-NE) or 0.963 (other directions); sigma is the workbook's S.D. rounded to 0.01 times that factor, so it moves in steps of 0.0094 / 0.0096. - [F] The tunnel did not run E, ESE, WSW or W; the field data cover all 16. The 5 m and 20 m planes exist at some points only (29 and 79 rows). - [F] One tunnel value (point 8, SW, 9 m: 0.474) has no counterpart in the workbook's results sheet, and the flat 'Sheet1' table has 0.63 there. - [F] Field anemometer heights above the tunnel floor are 13 - 21 m (3 - 11 m above ground by the readme's rule); the data paper says 3 - 9 m, so points 13 and 14 (11 m) disagree. - [F] The thermistor reads a SCALAR speed and the cup anemometer a 10-minute mean: a RANS |U| is biased low against both. - [F] The files do not say whether a wind direction names where the wind comes from. The cleaned tables read it as the direction the wind comes from (N = 0, E = 90). At NNW a solved field supports that reading: Eddy3D's OpenFOAM 12 run for the wind from NNW correlates R 0.935 with the tunnel's 9 m column labelled NNW and R 0.077 with its column labelled SSE, the opposite wind (clean/QUESTIONS.md QF9); the other directions are not solved and stay unverified. - [G] FIELD data: full-scale metres, no model, and nothing in m/s. The table carries U/UH (velocity_ratio) and k/UH^2 (k_ratio), UH = the mean speed at 7 m during each period; the approach-flow table carries U/Ub and k/Ub^2 (Ub = U at 9 m) in its U and k columns, which are dimensionless here. - [G] The readme lists the heights of the lee measurement as 1.5, 3.0, 6.0, 9.0 m; the table, the workbook, the drawing and the data paper have 1.5, 3, 4.5, 6 m. - [G] The tree belt is a porous canopy (shading rate 81 % after pruning): the served geometry is its 2 m × 74 m × 7 m envelope with the 1.2 m gap, and a solver needs a canopy (drag) model. - [G] The approach profile (28 Dec 2000, 120 min) and the lee measurements (26 Dec 2000, 10 min or more) are from different days. Behind the belt k/UH^2 is 0.040-0.072 at 5 H, 3-8 times below the approach flow's 0.26-0.31 at the same heights; the files do not say why. - [G] Only 28 numbers per quantity exist, on one centre section, one time-average each, with no repeats and no uncertainty. - [G] The wind direction relative to the belt normal during the measurement is not published; the case is two-dimensional by assumption. - [H] The table carries the legacy workbook's values (ratios at full precision x uH = 3.226 m/s), not RS_caseH.csv's: on the z/H = 1/2 and 1/4 planes (and in the approach profile) the CSV's U, V, W are the workbook ratios ROUNDED TO 0.01 before the scaling, a 0.032 m/s grid (+/-0.016 m/s). - [H] One row per instrument and point: velocity rows (split fibre) and concentration rows (hydrocarbon analyser) with the same point_id (the CSV's No.). The velocity line is x = 0.40 m and the concentration line x = 0.38 m, so 50 points have no velocity and 50 no concentration (-999.9 in the CSV); none of them is inside the building. - [H] V of the vertical-plane sheet (point numbers 1-220) is the exact negative of V at the same points in the horizontal-plane sheets (positive toward -y); it is multiplied by -1 so that V is positive toward +y everywhere, as in Cases A and D. - [H] The y = 0, z = 0.1 line (13 locations) is in two sheets and appears twice in the table, as exact copies once V is flipped; they are not independent repeat measurements. - [H] Concentration is the analyser's reading: upwind of the building (where the true value is zero) it is +/-0.1 ppm, and 97 of 826 values are negative (minimum -0.83 ppm). Log-based statistics (geometric mean, geometric variance) need a floor. - [H] The published header misspells w_rms as 'w_rsm'; the workbook's inflow sheet labels z as 'mm' but holds metres. - [H] The profile's uw and uv are published but the approach-flow table has no column for them. - [I] V on the y = 0 plane (53 of 91 points) is 0 in the CSV and 1e-9 ... 4e-6 m/s in the workbook, a symmetry value. It is NaN in the table; V is measured only at the 38 off-plane points. - [I] The table carries the legacy workbook's values; the CSVs round to 3 decimals (velocity, k) and 1 decimal (concentration, ppm). Every concentration is a multiple of 0.1 ppm: a 0.4 ppm value carries +/-12 % of quantisation. - [I] The paper's c0 = 39.2 ppm needs the ETHANE flow (10 % of the 400 cc/min, 6.67e-7 m3/s); its sentence 'ethane released at 6.67e-6 m3/s' is the mixture. With the mixture flow c0 would be 392 ppm. - [I] The workbook normalises the approach flow with 1.697 m/s and the points with 1.7 m/s (0.18 % apart). - [I] AF_caseI.csv ends in 53 rows of ',,' and two stray cells, '0.5*' and '1.5*' (chart axis labels): a reader that does not test z for a number takes them for data. - [I] Thirteen ground-centreline locations (x, 0, 0) are measured twice (they are on both the y = 0 and the z = 0 plane) with values up to 8.5 % apart; both are kept. The four y = 0, z = 0.05 wind points that appear twice in the workbook are identical copies and appear once. - [I] Only the TKE is published, no variances and no concentration fluctuation: k cannot be checked against the components. - [J] Temperature is in KELVIN in all three published CSVs; the API serves degrees Celsius (T) and kelvin differences (T_rms, heat fluxes). - [J] The published CSV is the legacy workbook's non-dimensional values rounded to 3-4 decimals and scaled back; the table is rebuilt from the workbook. The CSV's heat fluxes sit on a 0.0049 K m/s grid and k on 0.00021 m^2/s^2. - [J] c0 = 12.283 ppm includes the 5 % ethylene fraction of the released gas (workbook cell Co); q / (U_H H^2) alone is 245.7 ppm. The data paper states no c0 for Case J. - [J] RS_caseJ.csv lists the location (-0.16, 0, 0.004) twice (No. 1 and 153, identical values): a score over the table counts that point twice. - [J] The workbook's is not always the square of its sigma_u (up to 3 % in single rows); the table uses the variances, which reproduce k. - [J] The workbook lists an expanded uncertainty beside every quantity (no column here; clean/manifest.json cases.J.extra_measurements.expanded_uncertainty). The mean concentration is below its own uncertainty (about 1 ppm) at 42 % of the rows, and 193 uncertainty cells of the w'theta' flux are an Excel fill series, not values. - [J] The empty-tunnel profile changes along the tunnel: U at z = 0.019 H is 0.33 U_H at x/H = -2.5 (the published AF_caseJ.csv), 0.58 at the building's own position. The workbook has all three stations (clean/cases/J/approach_flow_full.csv). - [J] The +y and -y halves are not mirror images: over the 97 mirror pairs of the z/H = 0.25 plane U at +y is 11 % lower (means 0.558 and 0.629 U_H, R between the halves 0.955), so a y-symmetric model has a ceiling on this file. - [J] The readme's Reynolds number (43,000) is Case H's; U_H H / nu = 1.6e4 as the data paper states. - [J] uv, uw, vw are not published at the points; uc, vc, wc were not measured. - [K] The legacy workbook normalises with H = 0.2 m (Case H's building), UH = 4.94009 m/s and C0 = 18.218 ppm, not the cube's H = 0.06 m and U_H = 2.33 m/s; its normalised columns are 23.6 times smaller in concentration. The API's concentration_ratio uses the cube's c0 = 429.18 ppm. - [K] RS_caseK.csv has no point numbers; point_id is the row order of the file. - [K] The published CSV is the workbook rounded (3 decimals, concentrations 2), with k recomputed from the rounded rms values; the table is rebuilt from the workbook. - [K] AF_caseK.csv is Case H's approach-flow file, row for row (x = -0.493 m, tabulated for a 0.2 m building); it gives 2.24 m/s at z = 0.06 m against the stated U_H = 2.33. - [K] The lateral velocity V is dominated by a cross-flow offset (about 0.06 m/s, also in the empty tunnel); the +y and -y halves cannot confirm its sign. - [K] The mean concentration is not mirror-symmetric (the +y and -y values of one point differ by 44 % on average), so the mirror pairs are not a repeatability estimate; velocity and k are (3 % and 5 %). - [K] The workbook's / sigma_usc ('velocity magnitude', never below |U|) is not in the table. - [K] Data paper section 3.11 describes the measurements as 'approximately 30 points at eight heights downwind of the line source'; Case K has 203 points at five heights and a point source. - [K] Workbook units: x and y in millimetres in both data sheets; z in millimetres in 'Measurement data' but metres in 'Inflow'. - [L] RS_caseL.csv's temperature column is headed 'T(K)' but holds DEGREES CELSIUS (12.85 .. 29.86); AF_caseL.csv's 'T(K)' is kelvin (283 .. 295). The API serves degrees Celsius for both. - [L] The published CSV is the workbook rounded to 2-4 decimals; the table is rebuilt from the workbook. Its 'tc' column (temperature-concentration covariance, K ppm) and the workbook's 'Usc' (the vector magnitude of the mean velocity) are not in the table. - [L] BD_caseL.stl: the two bottom triangles (z = 0) of each of the 126 cubes (252 triangles) are wound inward; the served geometry.stl is the 126 boxes, outward. - [L] The readme's file list names AF_caseK.csv and RS_caseK.csv; the files are AF_caseL.csv and RS_caseL.csv. - [L] The readme's 'building surface temperature T_b = 14.8 degC' is the approach-flow air temperature at z = H; the workbook gives the building surface as 42.2 degC. - [L] The line source is 0.037 m x 1.2 m in the readme and data paper but a 0.04 m x 1.02 m patch in BD_caseL.stl. - [L] The +y and -y halves are not mirror images: the same point differs in U by 26 %, in k by 23 %, in C by 36 % on average, and U is 0.11 m/s lower and T 1.3 K higher at +y at z = 0.15 m (0.09 m above the roofs), so the tunnel flow has a lateral gradient; do not assume symmetry. - [L] The workbook carries no expanded uncertainty (the data paper mentions one for Case J only). - [L] AF_caseL.csv lists the profile from the top (z = 0.32 m) down and has no k. - [L] The measured tracer flux through the cross-sections equals q / 1.2 m to 0.86 .. 1.0 (street and cube-column lines averaged), which fixes q and the unit ppm; the cube-column line alone carries only 0.2 .. 0.4 of it. - [M] Field concentrations are in ppq (1e-15 volume fraction) and the tunnel's in ppm: 1 ppq = 1e-9 ppm. The cleaned table converts the field's to ppm and carries C / c0 for both, with the run's own c0 in the field. - [M] Every field run has its own reference speed U_R, mean direction WD_R and tracer flow q (clean/manifest.json -> cases.M.extra_measurements.field_runs); the row's wind_direction_deg is WD_R. - [M] The tunnel measured velocity only at 180°; at 175° and 185° the CSV has -999.9 in all seven velocity columns. - [M] RS-WT_caseM.csv is the legacy workbook rounded (2 dp velocity, 1 dp concentration), and its k is half the sum of the 2-dp RMS values, up to 0.005 m2/s2 from the workbook's; the cleaned table carries the workbook's values. - [M] The tunnel source is on the ground in the workbook, LC_caseM.csv and the STL (readme: 'on a rooftop'); the field source was 1.5 m above ground. - [M] H_R = 41 m is the anemometer height above the ground; the main roof under the reference point is about 33 m above it. c0 uses H_R = 0.068 m (41 m / 600 = 0.06833 m would give 131.2 ppm, -1 %). - [M] The field has an ambient floor of 8 - 11 ppq at points 11 - 15 in every run (not subtracted); the tunnel's concentration has a zero offset (two negative means at point 12) and an rms floor of 7 ppm. - [M] Only 11 of the 18 field runs are within 5° of 180°, the tunnel's direction; runs 1, 2 and 9 have U_R of 1.4 - 1.8 m/s. - [M] BD_caseM.stl is MODEL scale, wound inward on its ground and roof facets, and has 21 boundary and 7 non-manifold edges (7 on the rim of the panel, 14 on 7 slivers hanging on building roofs); the served geometry.stl is multiplied by 600, re-wound outward, without the slivers and closed with a base slab down to z = -1 m. - [M] The approach-flow heights are above the empty tunnel floor: at the profile station the model's taper is 0.052 m high, above the 10 lowest profile heights. ## Validation method Each case is solved through Eddy3D's production case writer (build-case -> run -> probe); the probe sits at exactly the tunnel's measurement coordinates. Scores are normalised by U_H (k by U_H^2; Case E by U at 15.9 m); hit rate uses D = 0.25, W = 0.05 in those units. FAC2 is harsh by construction on sign-changing V and W, where R and hit rate are the usable scores. The lateral sign of Case B is corrected before scoring. Cases C-E measure an omnidirectional SCALAR speed (thermistor): the model is scored as sqrt(|U|^2 + 2k) and as |U|, the two ends of the bracket the measurement must fall in. The ML surrogate predicts |U|/U_ref at 1.8 m; each case is scaled so its measurement plane lands there. On the full prediction window it does not respond to the wind direction at all (reversing the wind changes the field by < 1e-4 U_ref), so its scores grade the geometry it was given. ## Cite - Kikumoto, H., Okaze, T., Yoshie, R., Tachibana, T., Ishihara, T., Nonomura, Y., Kiyota, N., Kondo, H., Mochida, A., Tominaga, Y. (2026). Comprehensive Experimental Database for Validating CFD Simulations in Urban Wind Environment: Benchmark Cases Curated by the Architectural Institute of Japan. Japan Architectural Review 9(1), e70083. https://doi.org/10.1002/2475-8876.70083 - Tominaga, Y., Mochida, A., Yoshie, R., Kataoka, H., Nozu, T., Yoshikawa, M., Shirasawa, T. (2008). AIJ guidelines for practical applications of CFD to pedestrian wind environment around buildings. J. Wind Eng. Ind. Aerodyn. 96, 1749-1761. https://doi.org/10.1016/j.jweia.2008.02.058 - Each case's own requested citation and Zenodo DOI are in cases/{id}.json. - Cite the API: cite.api in index.json