Meta SustainableConcrete → Open3DCP
How an openly-licensed, carbon-aware kg/m³ table — released by Meta with UIUC and Amrize — re-formats into Open3DCP, lighting up the embodied-carbon and measurement-uncertainty groups that the UCI benchmark leaves empty.
1 · Native structure
Storage medium: a single flat table (.csv), one row per
mix · curing age, every constituent in kg/m³ of a 1 m³ batch. 149 mixes
/ 727 rows in the source; a 28-row, six-mix excerpt is committed here.
| Source field (data class) | Unit | Role |
|---|---|---|
| Cement | kg/m³ | binder |
| Fly Ash · Slag | kg/m³ | SCM |
| Water | kg/m³ | liquid (closes the wet-mix total) |
| HRWR | kg/m³ | admixture (high-range water reducer) |
| Fine · Coarse Aggregate | kg/m³ | aggregate (0 coarse ⇒ mortar) |
| Time | day | curing age |
| Temp | °C | curing temperature |
| Strength (Mean / Std) · # of measurements | psi · — | result + uncertainty (n) |
| GWP | kg CO₂/m³ | embodied carbon (cradle-to-gate) |
2 · Re-formatting into Open3DCP
embodied_carbon_kg_co2_m3 and the reported mean/std/n populate the measurement-uncertainty columns.Field map (crosswalk)
| Source field | Transform | Open3DCP column | Fidelity |
|---|---|---|---|
| Cement | kg/m³ → mass-% | cement_type_1 | type assumed (Type I) |
| Fly Ash · Slag | kg/m³ → mass-% | fly_ash · slag | exact |
| Water | kg/m³ → mass-% (+ w/b derived) | water · w_b_ratio | exact (closes wet mass) |
| HRWR | as-delivered → solids mass-% | superplasticizer | solids fraction assumed |
| Fine · Coarse Aggregate | kg/m³ → mass-% | concrete_sand · agg_size_57 | fineness/size assumed |
| Strength (Mean / Std / n) | psi → MPa | compressive_strength_mpa (+ _stddev_mpa, n_specimens) | exact |
| GWP | identity | embodied_carbon_kg_co2_m3 | exact |
| Time · Temp | identity | test_age_days · curing_temperature_c | exact |
3 · Result in Open3DCP
open3dcp-ingest — every constituent reported in kg/m³, so the
kg/m³→mass-% projection is arithmetic; specimen geometry and test method are not reported by the source and stay NULL.
Full report: fidelity.md.
Digital-twin coverage
The carbon trade-off this makes legible (28-day rows)
Constituents as mass-% of total wet mix. The same schema row carries strength and embodied carbon, so a low-clinker ternary beats plain OPC on both. Full file: meta-sustainableconcrete.open3dcp.csv.
| mix | cement_type_1 | fly_ash | slag | w_b_ratio | strength_mpa | embodied_carbon (kg CO₂/m³) |
|---|---|---|---|---|---|---|
| plain OPC (Mix_84) | 24.5 | 0 | 0 | 0.40 | 65.4 ±1.9 | 521 |
| OPC + fly ash (Mix_103) | 14.4 | 9.6 | 0 | 0.50 | 46.1 ±1.7 | 319 |
| OPC + slag (Mix_121) | 9.0 | 0 | 8.9 | 0.35 | 61.1 ±3.3 | 204 |
| ternary (Mix_88) | 7.7 | 6.6 | 11.2 | 0.25 | 110.5 ±4.3 | 204 |
Plain OPC: 521 kg CO₂/m³ for ~65 MPa. The ternary: 204 kg CO₂/m³ for ~110 MPa — higher strength, ~60 % less embodied carbon. (One illustrative pair, not a carbon-drives-strength trend: the ternary also runs a lower w/b — 0.25 vs 0.40 — and a second mix in this same excerpt at the same ~204 kg CO₂/m³ reaches ~61 MPa.)
4 · How this was built
python build/extract.py /path/to/boxcrete_data.csv # selects 6 mixes; psi→MPa, in→mm
open3dcp-ingest convert build/meta-sustainableconcrete.csv --kind flat --out .
# → meta-sustainableconcrete.open3dcp.csv .unmapped.jsonl .fidelity.md .fidelity.jsonA sample, not a re-host. Source © Meta Platforms, Inc. and affiliates, UIUC, and Amrize, MIT License — see NOTICE. Get the full 727-row table from the repository.