Metamodeling-based geotechnical characterization of the shear strain dependency of shear modulus of clayey soils from an Italian database
Open Access
- 3 June 2026
- journal article
- Published by Springer Nature in Acta Geotechnica
- Vol. 21 (9) , 6115-6134
Abstract
The quantitative modeling of the dependency of shear modulus on shear strain is a central aspect of the characterization of cyclic and dynamic properties of soils in the nonlinear elastic domain for earthquake geotechnical engineering applications such as site response analysis. This study investigates the dependency of normalized shear modulus on shear strain for clayey soils using the UNIFI database, which contains results from laboratory tests performed on soils from Italian geotechnical sites. A hybrid intelligent metamodeling framework is proposed within the Group Method of Data Handling (GMDH) paradigm. The approach combines the COMBI algorithm for systematic identification of candidate model structures with the shuffled complex evolution (SCE) algorithm for global optimization of model parameters, resulting in a hybrid COMBI-SCE modeling framework. The modeling procedure includes dependency analysis for the identification of influential predictors, generation of candidate model structures through COMBI, and parameter optimization using the SCE algorithm Query ID="Q3" Text="Please check the edit made in the article title. The predictive capability of the resulting models was evaluated using independent training and testing datasets and multiple statistical goodness-of-fit indicators. The results indicate that the most influential predictors of normalized shear modulus degradation in the analyzed dataset include the overconsolidation ratio, plasticity index, initial void ratio, mean confining stress, and shear strain amplitude. The resulting COMBI-SCE metamodel provides a compact analytical representation capable of capturing the nonlinear relationship between shear strain and normalized shear modulus while maintaining physical interpretability. The study demonstrates the potential of hybrid intelligent modeling approaches for developing predictive relationships for dynamic soil behavior and highlights the importance of incorporating site-specific data when developing data-driven geotechnical models.Keywords
Funding Information
- Dipartimento della Protezione Civile, Presidenza del Consiglio dei Ministri (DPC-ReLUIS 2022-2024)
- Università degli Studi di Firenze
This publication has 34 references indexed in Scilit:
- Similarity characterization of geotechnical engineering sites under the site-specific data scenarioBulletin of Engineering Geology and the Environment, 2024
- A simplified procedure for the prediction of liquefaction‐induced settlement of offshore wind turbines supported by suction caisson foundation based on effective stress analyses and an ML‐based group method of data handlingEarthquake Engineering & Structural Dynamics, 2023
- Use of GMDH-type neural network to model the mechanical behavior of a cement-treated sandNeural Computing and Applications, 2021
- A GMDH Predictive Model to Predict Rock Material Strength Using Three Non-destructive TestsJournal of Nondestructive Evaluation, 2020
- SPT-based liquefaction assessment with a novel ensemble model based on GMDH-type neural networkArabian Journal of Geosciences, 2019
- A novel ensemble model based on GMDH-type neural network for the prediction of CPT-based soil liquefactionEnvironmental Earth Sciences, 2019
- Dynamic characterization of fine-grained soils in Central Italy by laboratory testingBulletin of Earthquake Engineering, 2019
- Fuzzy Inference System-Based for TBM Field Penetration Index Estimation in Rock MassGeotechnical and Geological Engineering, 2018
- Uplift capacity prediction of suction caisson in clay using a hybrid intelligence method (GMDH-HS)Applied Ocean Research, 2016
- Applied Predictive Modeling2013