Stability assessment of the Three-Gorges Dam foundation, China, using physical and numerical modeling—Part I: physical model tests

Abstract

Foundation stability is one of the most important factors influencing the safety of a concrete dam and has been one of the key technical problems in the design of the Three-Gorges Project. The major difficulties lie in two facts. The first one is that the dam foundation consists of rock blocks, with joints and so-called ‘rock bridges’ and the gently dipping joints play a critical role in the foundation stability against sliding. The second one is that, even in the regions where the gently dipping fractures are most developed, there are no through-going sliding paths in the rock mass due to the existence of the rock bridges; so the dam could slide only if some of the rock bridges fail, so as to create at least one through-going sliding path. To date, due to unavoidable shortcomings in physical and numerical modeling techniques, there is not a single satisfactory method to solve the problem completely. For this reason, the integration of multiple methods was adopted in this study and proved to be an effective and reliable approach.

This Part I paper describes work based on the results of geological investigations and mechanical tests, relating to the geological and geomechanical models of the Three-Gorges Dam, and then a systematic study procedure was developed to carry out the stability assessment project. Then, 2D and 3D physical model tests for some critical dam sections were performed. In the physical tests, based on similarity theory, various testing materials were selected to simulate the rock, concrete, fracture and rock bridge. The loading and boundary conditions were also modeled to meet the similarity requirements. The failure mechanism was derived through a progressive overloading that simulated the upstream hydrostatic pressure applied to the dam, and the factor of safety was defined as the ratio between the maximum external load inducing the start of sliding instability of the dam foundation and the upstream hydrostatic load. The experimental results indicated that the stability of the Three-Gorges Dam foundation satisfies the safety requirements. Nevertheless, further discussions demonstrated that because of the incomplete definition of factor of safety adopted in the physical model tests, it is also essential to study the stability of the Three-Gorges Dam foundation using numerical modeling, which will be presented in the companion Part II paper.

Introduction

Foundation stability is one of the most important factors influencing safety of a dam. Instability of a dam generally results from pre-existing geological features in the foundation, such as faults, joints, soft rocks and solution channels, etc. Therefore, for a dam foundation, stability analyses and assessment are essential and form a critical part of the safety assessment.

The Three-Gorges water conservancy complex located about half way along the Yangtze River is the largest multipurpose water conservancy project ever built in China, indeed in the world (Fig. 1). The Three-Gorges Dam is one of the three main parts of this project (the others are the power houses and navigation facilities). It is a concrete gravity dam with a maximum height of 185 m and is designed to withstand a normal pool level of 175 m. According to the general layout of the dam (Fig. 2, Fig. 3, Fig. 4), 23 spillway-dam sections are located in the middle of the riverbed with a total length of 483 m, 26 powerhouse-dam sections are situated at the two sides of the spillway dam and have a total length of 1228 m. Along the foundation line, the dam sections are built on different ground levels, low on the river bottom and high on the riverbank. Detailed geological investigations show that although the dam foundation mainly comprises plagioclase granite that is intact, homogeneous, of low permeability and high strength, there also exist weathered zones, faults and joints in the rock mass. In particular, gently dipping joints are well developed locally. When the project is finished and runs at the normal pool level, the resulting reservoir will extend nearly 600 km upstream and have a total storage capacity of 39.3 billion cubic meters, which is of extreme importance to the safety of the dam.

During the past decades, the Yangtze Water Resources Commission (CWRC), with the collaboration of the Chinese Academy of Science, the Ministry of Geology, and a number of universities and institutes of China, carried out extensive geological investigations, field and laboratory testing of the rock and soil properties related to the Three-Gorges Dam. This research work has been thorough and systematic, and provided sufficient data for the stability analyses [1]. However, to study and assess the stability of the Three-Gorges Dam foundation, the following questions have to be addressed:

• The dam foundation mainly consists of rock blocks, various faults, fractures, joints and so-called ‘rock bridges’ (defined as the intact rock between the ends of sub-parallel fractures). Most of the faults and joints have a steep dip angle, but there are gently downstream-dipping joints that are actually the most important factors influencing the stability of the dam. On the other hand, even in the regions where the gently dipping fractures are most developed, no deterministic and through-going sliding paths in the rock mass exist due to the presence of the rock bridges. Thus, the dam could slide only if some of the rock bridges fail, so as to create at least one through-going sliding path. Therefore, in the stability analyses, taking account of the effects of the rock bridges becomes one of the key issues.

• Generally, two types of modeling approaches are adopted in stability analyses: physical model test and numerical modeling, including limit equilibrium method (LEM), finite element method (FEM), distinct element method (DEM) and discontinuous deformation analyses (DDA). The LEM has the advantage of simplicity and can provide a simple index of relative stability in the form of the factor of safety. Nevertheless, it considers only force and moment equilibria; deformations and material constitutive relations are not accounted for [2], [3]. The FEM considers deformations and material constitutive relations and provides more information about distributions of stresses, deformations and possible failure or damage zones, but has limited capacity in identifying explicitly the critical state of sliding and failure of the dam. The DEM/DDA approach is based on dynamic motion of rock blocks that can also be deformable [4]. One of the strengths of the DEM/DDA method is that, as time progresses, the blocks are allowed to move and deform, so that the mode of failure becomes apparent [2]. However, although this approach is specifically suited for carrying out discontinuum analyses in rock mass, it suffers some practical limitations since numerical stability and convergence depend on the proper selection of time steps and damping parameters [5]. More importantly, the discrete numerical approaches, like DDA and DEM, needs detailed fracture system geometry that is often not readily available [6], and they are not efficient for problems with non-persistent fracture systems and rock bridges. The physical model tests that use simulated materials based on the theory of similarity can provide a direct perceptional methodology [7], [8] and can automatically simulate the system performance of both dam and its foundation from elastic and elastic–plastic deformation to failure during the loading process [9], [10], [11]. However, such modeling costs much more and offers a limited amount of experimental output. Due to these shortcomings in each of the methods stated above, there is no single satisfactory method to fully solve the problem and so the integration of multiple methods became necessary. The main concern is how to properly treat the differences among their results and reduce the impacts of their shortcomings in the assessment of the stability of the dam.

• The Three-Gorges Dam consists of more than 50 dam sections with transverse concrete joints connecting them. The stability conditions are different between the sections because of the differences in the local geological conditions. Detailed study of each dam section is too costly and may not be necessary. Hence, picking out the representative sections for stability studies, using the knowledge from the geological investigation and conceptualization [12], [13], [14], is a critical process.

A systematic study procedure was established to carry out this stability assessment project. Based on the results of geological investigations and mechanical tests of the rock mass, the geological and geomechanical models of the Three-Gorges Dam foundation were firstly established and a study procedure for the stability assessment was provided. Then, using the same geomechanical model, 2D and 3D physical model tests for some critical dam sections were performed. In the physical model tests, various testing materials were selected to simulate the rock, concrete, fracture and rock bridge. The prototype load and boundary conditions were also modeled according to similarity theory. During the testing process, the displacements at some important positions of the dam structure and foundation were monitored. The failure mechanism of the dam foundation was derived through a progressive overloading that simulated the upstream hydrostatic water pressure, and the factor of safety was defined as the ratio between the maximum external load inducing the start of sliding instability of the dam foundation and the upstream hydrostatic load applied to the dam. The comparisons between the results of 2D and 3D physical model tests were conducted.

This study mainly concerns the following aspects: (1) establishment of the geomechanical model of the dam foundation; (2) physical model tests; (3) numerical modeling; (4) comprehensive stability assessments based on modeling results; (5) additional treatment and reinforcement design. Part I of this paper describes the characterization and conceptualization of the geomechanical conditions of the foundation, and physical model tests. The Part II companion paper presents the numerical modeling, comprehensive assessment, reinforcement measures, and the overall conclusions.