Integrating cluster-based knowledge for local undrained shear strength prediction
Integrating cluster-based knowledge for local undrained shear strength prediction
The prediction of local (i.e., site-specific) undrained shear strength from soil properties (e.g., water content, Plasticity index, Liquid Limit) is a cost-effective solution to preliminary assess the proper foundation type at a project site. Nevertheless, the local dataset is often incomplete and sparse to infer robust prediction of local strength parameters. To bypass such inconvenience, global or soil-type correlations might be employed to predict undrained shear strength from local soil properties measurements. However, by doing so two main drawbacks can arise: a great uncertainty is introduced for prediction of undrained shear strength due to the large number of soil types the global database is usually composed. Secondly, a problem of pertinence might arise when soil type-specific correlations are employed. In this study the local predictions of undrained shear strength are estimated integrating clustered global information through a Bayesian formulation, which allow to assess the problem of local dataset incompleteness. The methodology is shown by making use of an already compiled and published database.
Stefano Collico; Giovanni Spagnoli; Alessandro Fraccica; Enrique Romero
18th European Conference on Soil Mechanics and Geotechnical Engineering (ECSMGE2024)
A - New developments on structural design
Contact Information
- Secretary General - Dr A M McNamara
- Email: [email protected]
- Phone: +44 20 7040 8154
- City, University of London
- Northampton Square
- London EC1V 0HB. United Kingdom