UC Berkeley

The assessment of tunneling- and deep excavation-induced structural damage is in great demand due to the fast development of underground spaces in major cities. Tunneling and deep excavations may cause substantial ground movements and surface building damage, and the potential damage needs to be assessed. The limited site characterization and building condition surveys and the lack of knowledge in soil-structure modeling may cause significant uncertainty in the building damage assessment. The substantial potential damage and uncertainty have made tunneling- and deep excavation-induced ground movements a human-made hazard in urban environments. To design resilient urban infrastructure, the consequence of this human-made hazard needs to be quantified, and a resilience performance-based engineering approach needs to be developed. This study borrows the performance-based engineering approach in natural hazard engineering, especially in earthquake engineering, and creates a probabilistic performance-based engineering (PPBE) framework for the assessment of building damage in large tunneling and deep excavation projects. This dissertation focuses on hazard analysis, soil-structural analysis, and some damage analysis methods in the proposed PPBE framework because they are significantly different from the existing natural hazard assessment frameworks. The other components of the PPBE, such as building asset inventory development and economic/environmental loss estimation, are briefly discussed and more theories can be referenced from other natural hazard assessment approaches.

A computational tool named Uncertainty Quantification in Tunnel and Excavation Soil-Structure Interaction (UQ-TESSI) is developed to implement the proposed hazard analysis, soil-structure analysis, and the associated uncertainty quantification. UQ-TESSI incorporates a suite of deterministic soil-structure analysis models named Analysis of Structure Response to Excavation (ASRE) with the PPBE framework. Applying the proposed PPBE framework in the assessment of tunneling- and deep excavation-induced building damage faces two major challenges: (1) the large number of buildings needed to be analyzed, (2) the high-dimensional uncertainty associated with material, workmanship, and tunneling- and deep excavation-induced soil-structure interaction (T&DE SSI) modeling processes. These challenges are alleviated by: 1) creating T&DE SSI modeling and computation tools that minimize uncertainty with reasonably small modeling effort, 2) creating probabilistic models to quantify the uncertainty in the assessment approach, and 3) applying high-performance computing and advanced uncertainty propagation and quantification techniques. The application of the proposed PPBE approach and UQ-TESSI are demonstrated with multiple case studies enabling building performance assessment in T&DE on a regional or community-level scale.

The case study results enable the following observations. The application of the proposed PPBE and 2D soil-structure models confirmed several widely recognized soil-structure interaction mechanisms. The 2D modeling case studies also revealed the large uncertainty associated with the building stiffness estimation and the inability of 2D models to appropriately simulate the effects of analyzing progressive excavations. For the 3D case studies, soil-structure analysis modeling was found to successfully capture the out-of-plane spatial variability in both tunneling and deep excavations. In addition, the coupling effect among masonry facade walls needs to be considered in tunneling, and the results indicate that neglecting the coupling effect may lead to an overestimation of building strain when the building is skewed compared to the tunnel axes. The case history data collected in the construction of the Elizabeth Line and several deep excavation projects in Norway confirmed the expectation that the ground movements in tunneling and deep excavation have strong spatial variability, and the spatial variability needs to be considered in the probabilistic modeling of ground movements in regional building damage assessment. Overall, it is observed that the uncertainty in ground displacement estimation in the current analysis approach produced the most uncertainty in the large-scale building damage assessment, and quantifyingthe uncertainty is a high-dimensional uncertainty quantification problem, which is arguably only feasible with high-performance computation and advanced Monte Carlo methods.