Preliminary automated methods of constitutive model parameters of carbonate sands in engineering practice

1. Home
2. Publications
3. Online Library
4. Preliminary automated methods of constitutive model parameters of carbonate sands in engineering practice

The calibration of constitutive model parameters from laboratory measurements is a crucial phase in geo-structure design. The selection of the most suitable constitutive model is often constrained by software capabilities and laboratory information rather than solely by engineers' preferences. When dealing with complex soil behavior, simplified constitutive models commonly used in practice (e.g., Mohr-Coulomb, Hardening Strain, and Hardening Small Strain models) may be insufficient. Depending on the software employed, more advanced constitutive models may require a greater number of geotechnical parameters to be calibrated. Consequently, the calibration process can be time-consuming, necessitating additional effort to assess the sensitivity of parameters to soil response. This paper introduces two novel workflows for automating the calibration of constitutive model parameters by integrating multiple laboratory measurements. Such workflows are easily applicable to different software platforms. We propose a software-based approach and a surrogate-based approach, both implemented under a Bayesian framework. The former method aims to provide a flexible tool for practitioners within a remote scripting environment and built-in software components for probabilistic prediction of constitutive parameters. The latter offers a more computationally efficient and flexible approach to handle a large number of simulations. The proposed methods are tested to calibrate the NorSand constitutive model parameters from triaxial and direct shear tests performed on Dogs Bay carbonate sands at relatively low confining stress. The dependence of constitutive model parameters on particle breakage is beyond the scope of this study and is left for future research. 

5th International Symposium on Frontiers in Offshore Geotechnics (ISFOG2025)

5 - Data Analytics and Machine Learning