High-fidelity data-driven multi-objective design of class F fly ash–geopolymer concrete
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Class F fly-ash geopolymers (FA-GPCs) can slash the CO₂ burden of structural concrete, yet their mix design still relies on costly trial-and-error. The first geometry-normalized database of FA-only geopolymer concretes was assembled—799 mixes from 67 publications, every reported strength converted to the 150 Ø 300 mm cylinder equivalent—and interrogate it with a three-scenario, data-availability workflow. Eighteen machine-, deep- and hybrid-learning models are benchmarked; a Voting ensemble predicts 7–28 days compressive strength with an R 2 of 0.962, while an LSTM + XGBoost hybrid forecasts mix-specific CO₂ footprints at an R 2 of 0.996. SHAP analysis reveals that the Si/Al ratio and initial curing temperature dominate early-age strength, whereas total Na₂SiO₃/NaOH and Na₂O/binder ratios control embodied carbon, underscoring an inherent strength–footprint trade-off. These surrogate models feed a multi-objective optimizer (NSGA-II, confirmed superior to NSGA-III and MOEA/D) that explores the 1st–99th-percentile bounds of 11 mix variables. The resulting Pareto fronts deliver eco-efficient recipes reaching 65 MPa at ≤30 kg CO₂/m 3 (at 90-d curing age) and extend to 87 MPa under relaxed carbon targets. Compared with state-of-the-art optimization that rely on mixed precursors or un-normalized data, the proposed pipeline widens the design space, lifts strength by 10–20 MPa and trims CO₂ by up to 70 %. The framework— in this order: data curation, geometry correction, scenario-specific ML, evolutionary optimization—cuts trial batching efforts and offers practitioners ready-to-deploy, class F FA-GPC mixtures while providing a transparent template for other alkali-activated binders which was confirmed by an experimental validation. • Geometry-normalized database of ~800 FA-geopolymer mixes enables apples-to-apples ML. • Voting ensemble predicts 7–28-d strength at R 2 = 0.962; LSTM + XGBoost CO₂ at R 2 = 0.996. • SHAP shows Si/Al ratio and curing-T drive strength, while Na-metrics govern CO₂ footprint. • NSGA-II yields Pareto mixes hitting 65 MPa at ≤30 kg CO₂/m 3 (90-d curing). • Framework cuts trial batches and offers ready-to-use class F FA-GPC recipes.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it