Predicting 28-Day Compressive Strength of Concrete from Field Data: Uncertainty Quantification and Seasonal Interaction Effects

Abstract Compressive strength at 28 days remains the standard criterion for concrete acceptance, yet it is unknown until four weeks after placement, well after decisions such as formwork removal must already be made. Most predictive models in the literature are developed and validated on laboratory-mixed specimens, leaving open how well they generalize to field-collected data, where records are incomplete and the early-age conditions of the specimens vary with site and season. This study addresses that gap using 1,018 field records from construction projects in Mazandaran province (2023–2025), evaluated under two realistic scenarios: with and without an early-age (7-day) strength measurement. A stacking ensemble combining XGBoost, SVR, and MLP was benchmarked against several meta-learner variants; formal statistical testing (Wilcoxon signed-rank and Diebold–Mariano tests) showed no significant improvement from more elaborate meta-learners over a simple linear combination once base learners were properly tuned. Conformal prediction was therefore used to quantify predictive uncertainty directly: interval widths nearly doubled in the absence of the 7-day test (11.5 versus 22.4 MPa at 90% confidence and 14.8 versus 27.1 MPa at 95%, standard CV+), with direct implications for quality-control decisions. Segment-wise error analysis showed that errors were largest at the extremes of the strength range when the 7-day test was unavailable, with no detectable seasonal difference, while SHAP interaction analysis revealed a season-dependent modulation of both early-age strength and slump effects that is physically consistent with the temperature sensitivity of cement hydration, although temperature itself was not measured.

Authors

Institutions

Publication Details

Journal
Research Square
Published
2026-11-17
DOI
https://doi.org/10.21203/rs.3.rs-11442269/v1
Primary Topic
Concrete and Cement Materials Research
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Predicting 28-Day Compressive Strength of Concrete from Field Data: Uncertainty Quantification and Seasonal Interaction Effects

Hamed Enayati, Mousa Nazari, Mahdi Kiani
3 links
Research Square
Concrete and Cement Materials Research
preprint

Predicting 28-Day Compressive Strength of Concrete from Field Data: Uncertainty Quantification and Seasonal Interaction Effects

Hamed Enayati, Mousa Nazari, Mahdi Kiani
preprint en

Abstract

Abstract Compressive strength at 28 days remains the standard criterion for concrete acceptance, yet it is unknown until four weeks after placement, well after decisions such as formwork removal must already be made. Most predictive models in the literature are developed and validated on laboratory-mixed specimens, leaving open how well they generalize to field-collected data, where records are incomplete and the early-age conditions of the specimens vary with site and season. This study addresses that gap using 1,018 field records from construction projects in Mazandaran province (2023–2025), evaluated under two realistic scenarios: with and without an early-age (7-day) strength measurement. A stacking ensemble combining XGBoost, SVR, and MLP was benchmarked against several meta-learner variants; formal statistical testing (Wilcoxon signed-rank and Diebold–Mariano tests) showed no significant improvement from more elaborate meta-learners over a simple linear combination once base learners were properly tuned. Conformal prediction was therefore used to quantify predictive uncertainty directly: interval widths nearly doubled in the absence of the 7-day test (11.5 versus 22.4 MPa at 90% confidence and 14.8 versus 27.1 MPa at 95%, standard CV+), with direct implications for quality-control decisions. Segment-wise error analysis showed that errors were largest at the extremes of the strength range when the 7-day test was unavailable, with no detectable seasonal difference, while SHAP interaction analysis revealed a season-dependent modulation of both early-age strength and slump effects that is physically consistent with the temperature sensitivity of cement hydration, although temperature itself was not measured.

Research Square
Mazandaran University of Science and Technology (IR), University of Science and Technology of Mazandaran (IR)
Concrete and Cement Materials Research
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon