Answers are grounded in this paper and cite their sources.
This study develops and validates a data-driven model based on an artificial neural network (ANN) to predict the effective shear strength parameters (effective angle of internal friction and effective cohesion) of granular soils from basic index properties.A database of 1,105 secondary records, obtained from standardized geotechnical tests performed at Peruvian universities, was used for training and internal validation.External verification was carried out with 39 independent samples tested by consolidated-drained direct shear in accordance with ASTM International D3080/D3080M and Peruvian Technical Standard (Norma Técnica Peruana, NTP) 339.171.The network, with three hidden layers, dropout, and batch normalization, uses particle size distribution, Atterberg limits, dry density, and moisture content as input variables.In the external test, the model achieved a coefficient of determination (R 2 ) of 0.86 for the effective angle of internal friction (mean absolute error [MAE] = 1.21°; root mean square error [RMSE] = 1.75°) and 0.77 for the effective cohesion (MAE = 1.09 kPa; RMSE = 2.14 kPa).Unlike previous studies based on limited or heterogeneous datasets, this work uses a broad database of Peruvian soils, which facilitates regional calibration.Gradient and sensitivity analyses indicated that particle size distribution is the most influential factor on shear strength, whereas moisture content has a comparatively smaller influence.The results suggest that the proposed network may serve as a preliminary estimation tool for the shear strength of Peruvian granular soils when basic index properties are available.However, given the small external test set and the regional scope of the data, its extension to other soil types or regions requires further validation.
No takes yet. Share an insight, caveat, or question.
Alarcón et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: