Sand Triaxial Test Database: Drained monotonic triaxial compression tests on granular materials

This dataset is a curated database of drained monotonic triaxial compression tests on granular soils, compiled from published literature and open datasets into a single, uniform, machine-readable format. It covers 59 materials — 46 natural and industrial sands (including seven artificially prepared grading fractions of a fluvial sand and a calcareous coral sand), six artificially graded coarse-grained soils, one rockfill, and six glass-bead materials (including mixtures of angular and rounded particles) — tested at 27 institutions worldwide, comprising 381 tests — mostly isotropically and normally consolidated, with a subset of anisotropically (K0) consolidated and overconsolidated tests — with approximately 80,000 digitized measurement records of axial strain, volumetric strain, and effective radial and vertical stress.
The materials span a wide range of gradations (median grain size d50 from 0.10 to 7.2 mm, coefficient of uniformity Cu from 1.0 to 431) and the tests cover initial relative densities from very loose to very dense and effective confining stresses from 5 to 3500 kPa. Each material is stored as a single self-contained CSV file with six standardized sections: file information, full source references with DOIs, index properties (d50, Cu, Gs, e_max, e_min, critical-state friction angle), particle size distribution(s), a test programme catalogue (consolidation conditions, initial stresses, void ratio, relative density per test), and the stress–strain–volumetric measurement records. An index file summarizes all materials for browsing, a consolidated reference list links every cited source to the materials that use it, and a README documents the format, sign conventions, and ready-to-use Python code for reading the files. Empty cells consistently indicate values not reported in the original source; most test curves were digitized from published figures and carry the usual digitization tolerance.
The database is intended as a benchmark resource for the development, calibration, and validation of constitutive models for sand, for data-driven and machine-learning approaches to soil behaviour, and for comparative studies of monotonic sand response across gradations, densities, and stress levels. Users are asked to cite, alongside this database, the original data sources listed in each material file. The compilation draws on the cited publications, the GEOLAB Material Properties Database, and the SoilModels standard datasets.
This database is a compilation of previously published experimental work. Users are kindly asked to cite both this database and the original source publication(s) of each material they use; the sources are listed per material in the SOURCES section of each CSV file and compiled in references.csv.