Applying probabilistic approaches for reliable sequential excavation method tunnel design and construction

Zheng, Haotian

Publication

Applying probabilistic approaches for reliable sequential excavation method tunnel design and construction

Zheng, Haotian

Advisor

Mooney, Michael A.
Gutierrez, Marte S.

Date Issued

2022

Abstract

The sequential excavation method (SEM) is commonly employed in soft ground urban tunnel construction. Large-scale SEM in the midst of urban infrastructure is complex because allowable ground deformations are relatively small. Considerable uncertainties exist during SEM design and construction due to limited knowledge of ground conditions and flexible construction measures as allowed by contractual guidelines. The conventional practice using deterministic analysis to evaluate SEM-induced ground and structural responses does not account for the uncertainties, and therefore is less adaptive and robust, leaving the SEM design and construction blindly risk-taking. Given the considerable technical and financial risks taken by SEM projects, it is imperative to develop efficient and accurate probabilistic approaches to help designers and contractors make well-grounded decisions based on updated knowledge of risk along the SEM construction process. As the first contribution of the thesis, a comprehensive case study of the Regional Connector Transit Corridor (RCTC) cavern construction is conducted. The performance of the three-drift seven-stage urban SEM was assessed, and complex ground and structural behaviors were evaluated every excavation step by analyzing the field observations. The average total ground surface settlement is about 20 mm. The total volume loss of 0.45% is considered an excellent-quality practice considering the large scale and shallow overburden of the cavern. An automated program was developed to analyze four-dimensional deformation measurements in real time, providing prompt ground and structural behavior assessment to facilitate the decision-making process. Second, a three-dimensional (3D) finite-difference method (FDM) model is developed to simulate the SEM construction process and surrogate models are developed to accurately and efficiently capture the complex 3D FDM deformation responses induced by SEM tunneling. The efficacies of four surrogate methods were examined. A sensitivity analysis was performed to determine the most influential geotechnical parameters as the surrogate model inputs. A design of experiments was performed based on Sobol sequence sampling. The results indicate two Polynomial-Chaos-Kriging (PCK) models exhibit the best performance with the model-average normalized root mean square errors (NRMSE) less than 3%. Third, a surrogate-based Bayesian approach is developed to update uncertain ground parameters and deformation predictions during the RCTC SEM construction. The PCK surrogate model developed in the previous part was used to evaluate SEM-induced ground and structure deformations to be compared with the field observations. Time-series observations of multiple measurements are used to form the likelihood function. The posterior distributions derived from the affine invariant ensemble sampling method reveal higher elastic modulus and cohesion of the fresh Fernando formation than what was assumed before the construction. The uncertainties of the geotechnical parameters are substantially reduced, which translates into reduced risk simply because prediction uncertainty is far less. The updated parameters for all considered monitoring sections show similar results, indicating similar ground conditions along the cavern alignment. Lastly, a reliability approach is developed to evaluate ground and structural stabilities induced by SEM construction under the uncertainty of ground conditions and the variability of design measures. The variations of two geotechnical parameters, two shotcrete parameters, and three types of SEM toolbox items were considered in the reliability analysis. The influence of each parameter on the two reliability indices was quantified and discussed.

Rights

Copyright of the original work is retained by the author.

Applying probabilistic approaches for reliable sequential excavation method tunnel design and construction

Zheng, Haotian

Citations

Advisor

Editor

Date

Date Issued

Date Submitted

Keywords

Collections

Files

Research Projects

Organizational Units

Journal Issue

URI

Embargo Expires

Abstract

Associated Publications

Rights

Embedded videos