MARS inverse analysis of soil and wall properties for braced excavations in clays

A major concern in deep excavation project in soft clay deposits is the potential for adjacent buildings to be damaged as a result of the associated excessive ground movements. In order to accurately determine the wall deflections using a numerical procedure such as the finite element method, it is...

Full description

Saved in:
Bibliographic Details
Main Authors: Zhang, Wengang, Zhang, Runhong, Goh, Anthony Teck Chee
Other Authors: School of Civil and Environmental Engineering
Format: Article
Language:English
Published: 2019
Subjects:
Online Access:https://hdl.handle.net/10356/87216
http://hdl.handle.net/10220/49287
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
Description
Summary:A major concern in deep excavation project in soft clay deposits is the potential for adjacent buildings to be damaged as a result of the associated excessive ground movements. In order to accurately determine the wall deflections using a numerical procedure such as the finite element method, it is critical to use the correct soil parameters such as the stiffness/strength properties. This can be carried out by performing an inverse analysis using the measured wall deflections. This paper firstly presents the results of extensive plane strain finite element analyses of braced diaphragm walls to examine the influence of various parameters such as the excavation geometry, soil properties and wall stiffness on the wall deflections. Based on these results, a multivariate adaptive regression splines (MARS) model was developed for inverse parameter identification of the soil relative stiffness ratio. A second MARS model was also developed for inverse parameter estimation of the wall system stiffness, to enable designers to determine the appropriate wall size during the preliminary design phase. Soil relative stiffness ratios and system stiffness values derived via these two different MARS models were found to compare favourably with a number of field and published records.