Pavement Concrete Datasets for Compressive Strength and Chloride Ion Penetrability Prediction
Description
This dataset contains curated pavement-concrete mixture data assembled for the analysis and prediction of compressive strength and chloride ion penetration (CIP). It consists of two datasets: 1,430 pavement-concrete records for compressive strength and 246 records for chloride ion penetration. The data were compiled primarily from publicly available U.S. federal and state transportation-agency research reports, together with selected peer-reviewed studies. The original sources differed in reporting format, units, mixture-property ranges, and completeness. The records were therefore systematically extracted, screened, unit-harmonized, cleaned, and organized into standardized pavement-concrete datasets. The compressive-strength dataset includes mixture constituent quantities, supplementary cementitious materials, coarse and fine aggregate contents, testing age, slump, air content, and measured compressive strength in MPa. The chloride ion penetration dataset includes mixture constituent quantities, slag, fly ash, silica fume, metakaolin, fine and coarse aggregate contents, maximum aggregate size, chemical admixture quantities, slump, air content, and measured chloride ion penetration expressed as charge passed in coulombs. Data screening was performed to retain mixtures representative of the pavement-concrete domain. Records containing missing, erroneous, or physically inconsistent mixture or measured-property values were removed before final dataset preparation. The datasets were developed as part of the study entitled “Explainable Ensemble Machine Learning Framework for Pavement Concrete Compressive Strength and Chloride Ion Penetration Prediction” and are provided to support reproducibility, independent model development, comparison, and benchmarking of data-driven pavement-concrete property prediction methods.
Files
Steps to reproduce
The database was assembled from published pavement-concrete mixture and performance data, predominantly from U.S. federal and state transportation-agency technical reports and supplemented by selected peer-reviewed studies. Reported mixture proportions, fresh-concrete properties, aggregate characteristics, supplementary cementitious material contents, and measured performance properties were extracted and converted to consistent units. The compiled records were screened to retain mixtures representative of pavement concrete. Screening considered water-to-binder ratio, air content, slump, testing age, and aggregate contents. Records containing missing, erroneous, or physically inconsistent mixture or performance values were removed. The retained records were subsequently harmonized into common variable definitions and units. The resulting datasets contain 1,430 compressive-strength records and 246 chloride ion penetration records. No additional preprocessing, normalization, train-test splitting, or feature scaling has been applied to the deposited data. Users may perform such procedures according to the requirements of their intended analysis.