Damage probability assessment for adjoining buildings to deep excavations in soft soils

• Autores: Carlos Javier Sainea Vargas
• Directores de la Tesis: Mario Camilo Torres Suárez (dir. tes.)
• Lectura: En la Universidad Nacional de Colombia (UNAL) ( Colombia ) en 2019
• Idioma: español
• Enlaces
  • Tesis en acceso abierto en: repositorio.unal.edu.co
• Resumen

  • In densely populated cities, the lack of space on the surface gives rise to the need for utilizing subsoil through different infrastructures, in which deep open excavations are required. However, there are several challenges to face, related to local geotechnical conditions as in the presence of soft soils, or serviceability requirements as potential damages on neighboring buildings may occur. Apart from the unfavorable behavior of soft soils, given its low resistance and high compressibility, their properties are spatially variable. These aspects should be considered in the analyses when assessing potential damages in buildings caused by ground movements during the excavation.

    In this research, a combination of numerical and probabilistic methods is considered through the use of a random field-based finite element modeling of deep excavations in soft soils. Finite element modeling allows performing tridimensional analysis, simulating staged construction, and including complex soil behavior. Probabilistic methods are useful to address the uncertainty in constitutive parameters related to spatial variability, expressing the response of the models in terms of damage probabilities, and updating initial predictions using information from other sources. Constitutive models Hardening Soil and Hardening Soil Small Strain are considered to model soft soil behavior, and selected parameters E0ref and E50ref were represented as random variables and random fields with different anisotropy in numerical models. Damage probability assessment analyses in terms of damage probability indexes were performed for a synthetic excavation of idealized geometry in Bogotá soft soils, and a real excavation project in Mexico City soft soils.

    The obtained results indicate a significant increase in assessed probabilities when not considering soil stiffness at small strains. Slight to moderate effects were found when changing the simulated construction sequence, using random-variable based models or changing the anisotropy ratio in random field-based models, except in highly anisotropic random fields. Ground movements and building damage potential distribution depend on boundary conditions, and they are different when changing the constitutive model employed or the simulated construction sequence.

    Similar results were attained when using the response surface, or the point estimates methods. In the former method, polynomial equations were employed to approximate numerical model performance in order to assess damage probabilities, obtaining similar results when using Gaussian random variables or the distributions found in inferential analyses. The polynomial equation approximation was also useful to update initial damage probabilities employing information from semiempirical methods or a combination of these and actual measurements.