Automated Generation of Ground Model Incorporating Soil Spatial Variability

The field of geotechnics undergoes a significant transformation in solving complex engineering problems by adopting digital technologies and automation processes. Among many other improvements, digital ground modelling has become state of the art when it comes to infrastructure projects as one of the most comprehensive ways to represent geological and geotechnical conditions of soil and rocks. It also allows for integration of digital ground models with other subsystems, including underground infrastructure, utilities, above ground infrastructure, and construction processes. While the domain of 2D and 3D geological modelling has advanced significantly in the past decades, incorporating soil
spatial variability, and thus assessing the underground heterogeneity and parameter uncertainty, remains a challenge. In this regard, this paper proposes an automated framework for generating a digital ground model based on site-specific borehole data within a Building Information Modelling (BIM) framework.
Furthermore, the digital ground model incorporates geotechnical parameters’ spatial variability using Conditional Random Fields (CRFs). The soil parameter variability is simulated through CRF randomization of soil strength parameters to determine their values at the non-sampled points. Borehole-based ground modelling is conducted using GSTools and a BIM software environment for the case study of the New Belgrade area in Serbia. Integrating the above-mentioned data into one holistic digital ground model offers numerous advantages and leads to more reliable probabilistic-based predictive numerical models of structures embedded in variable strata.