Mochikoshi Tailings Dams, Dikes No. 1 AND No. 2, Japan (1978)

Mochikoshi Tailings refers to a tailings pond from a gold mine in Japan, which was impounded by three dams, all constructed using the upstream construction method Ishihara (1984). The dikes 1 and 2 failed downstream in the event of Izu-Ohshima-Kinkai earthquake (M = 7.0) of 1978 in Japan. The failure resulted in the release of large volumes of tailings (Ishihara 1984; Ishihara et a!. 1990). The dike No.1 failed during the earthquake, whereas the other dam, dike No. 2, failed almost 24 hours after the earthquake. This case has also become a classic example to examine the validity of numerical procedures (Olson, 2001).

Fig. 2.15b shows cross section of the dike No. 2, it also shows tailings surface before and after failure. As Ishihara (1984) reported, the starter dam was constructed in 1964 by placing soils of volcanic origin. The weight of the transporting bulldozers was thought to compact these soils. The resulting tailings deposits were highly stratified (3 - 7 cm thick) sandy silt and silt with about 80% of fines content with water content greater than liquid limit.

一

        5   ＼

The main shock of the earthquake occurred around noon. This was followed by smaller after shocks. Two after shocks of magnitudes 5.8 and 5.4 occurred the next morning around 7.30 a.m. eyewitness noticed cracks in the down stream slope around 8.30 a.m.

and the dike failed around 1.00 p.m. Ishihara (1984) argued that the delayed failure occurred because of the rise in phreatic surface due to pore water pressure migration and made an attempt to analyze it. The delay in the failure and the inhomogeneous layered nature of the tailings indicate an effect of contrast in permeability of sub-layers within the liquefied depth of sub-soil. Moreover, the stratification of sub-layers and their high water content explored by boreholes after the event suggest the possibility of seepage flow of pore water after the end of shaking in which the pore water pressure dissipation was associated with significant volumetric change and flow deformation within the liquefied sub-soil.

Fig.2.15. Cross section of Mochikoshi tailings dams, (a) Dike N°1, (b) Dike N°2 (Ishihara et al., 1990)

(a)

(b)

Chapter 2: Characteristic behavior of Sand and Liquefaction. 40

2-7-3. Calaveras Dam, California, USA (1918)

This case history is described in two articles in Engineering New Record (Hazen and Metcalf, 1918; Hazen, 1918) and a paper in the Transactions of the ASCE (Hazen, 1920). The Discussion that accompanies the Transaction paper is illuminating, and it appears that several other dams failed similarly to Calaveras. Calaveras Dam was completed as a 64 m (210 ft) high earthfill dam. It suffered a flow failure near the end of its construction which led to a redesign.

The original dam that failed was of un-compacted fill shells (steam shovels fill) which were used to contain additional hydraulic filling. The hydraulic fill was placed at the outside limits of the shell, so that soil settled out preferentially leaving relatively sandy shells and a very soft silt core. This core consolidated under its own weight, but a slower rate than that of further fill placement.

This scheme of construction was not uncommon at the end of the 19th and early 20^th centuries, and is illustrated schematically in Fig2.16. Even though a clear distinction is shown between toe (shell in modern parlance) and core, in reality this was somewhat gradational. The fill was primarily material taken from the surrounding hillsides. The steam shovel fill was broken up soft sandstone, although Hazen (1918) noted that it was not true sandstone as the broken rock decomposed into particles “almost as fine in grain size as clay”.

Fig.2.16. Sketch of Calaveras Dam failure showing surface before and after slip (Hazen, 1918).

The steam shovel fill was not compacted other than from traffic moving around its surface; an average in place bulk unit weight of 18.8 kN/m3 is reported. The hydraulic fill comprised both surface soil and disintegrated soft rock, and was not compacted. The specific gravity was noted as being unusually light with Gs≈2.3 rather than 2.6–2.7 associated with siliceous materials.

On 28 March 1918, about 610,000 m3 moved 90 m upstream while dropping 30 m in elevation as it did so, within a time of about five minutes. The post-failure slope angle was about 5.7˚.

It is interesting to mention that this case history is one of the earliest examples of a slope failure caused by liquefaction of sand and the failure was not associated with any earthquake shaking. Hazen (1920) recognized the process of liquefaction in his comments regarding the failure and it is thought that it was the first time; the description of the failure mechanism reveals the procedure of liquefaction without calling it “liquefaction”.

Chapter 2: Characteristic behavior of Sand and Liquefaction. 42

2-8. Summary and Concluding Remarks

This chapter presents a literature review of characteristics and behavior of sands with regard to the liquefaction problem based on data from laboratory element tests. The main factors related to the scope of this study were addressed. The main conclusions from this overview are as follows:

• Volume change in granular materials is coupled with applied shear strain which results in dilation or contraction.

• The amount of dilation increases with sand relative density.

• The stress-strain behavior of sand is controlled by its skeleton and the seepage flow conditions of the pore water. The volume change condition can vary in a wide range, and the undrained (constant volume) and the fully drained (free) conditions are just two special cases within this wide range. To capture the post-liquefaction response of an earth structure after the end of earthquake, a realistic numerical modeling is required to facilitate an accurate prediction of soil element behavior in different loading conditions.

The following are the main findings of this literature review of the case histories:

• from observations of case histories, liquefaction-induced flow slides have widely occurred in very gentle sloping ground conditions in past earthquakes.

• Most of these failures started during shaking or some time after the main shock of the earthquake motion from few seconds to days.

• Applying steady-state concepts in such cases to estimate the undrained residual strength of liquefied soils (at the pre-earthquake relative density) suggests that failure should not occur as it provides with over-estimation strengths.

• The back calculated residual strengths from case histories indicate strongly the significant effect of considering the volumetric change mechanism on flow deformation mechanism.

• The post-liquefaction strength seems not to be controlled by only the conditions before or during earthquake shaking, but to the flow condition of the liquefied sub-layers (e.g.

stratification of sub-layers, contrast in permeability).

• Field case histories do not have the information to prove whether seepage of pore water within sub-layers took place or not. However, the review of case histories of several cases where seepage of pore water and the subsequent volumetric change could have probably played a role in the failure highlights the need for study of soil expansion due to seepage along with the site investigation techniques to detect the characteristics of sub-layers (e.g.

thickness of sub-layer, permeability contrast).

44

CHAPTER 3

REVIEW OF THE CURRENT THEORIES ON

POST-LIQUEFACTION FLOW DEFORMATION ANALYSIS OF GROUND

3-1. Introduction

Engineering estimation of the deformations and displacements likely to occur as a result of liquefaction or pore-pressure-induced ground softening is a difficult and very challenging step in most projects, and this is an area where further advances are needed.

The behavior of a softened material after liquefaction onset is called post-liquefaction response. The post liquefaction strength is important as the flow slide in liquefied grounds usually takes place after shaking. Extensive attempts have been made to describe the post liquefaction strength of sands, based on the various concepts and testing conditions (e.g. undrained or drained condition). In this chapter, a number of the most recent theories on post-liquefaction flow deformation analysis of ground based on physical model tests and/or numerical studies are presented as follows:

3-2. The Current Theories for Evaluating the Residual Undrained Shear Strength of Liquefiable Soil

Although it is clear that the majority of the deformation of the subsoil layer is generally caused by the decrease of its shear strength due to the liquefaction, however, it is not yet fully clear how the deformation progresses during the actual ground shaking.

Several research investigations are being conducted to clarify the mechanical properties of soil in the post-liquefaction state, but an agreement on how to treat liquefied soil for analyzing its deformation has not yet been reached.

Two different constitutive models of liquefied soil are used in practice. One model treats the liquefied soil as a viscous liquid, and the other mode as a softened solid. Those numerical techniques can be divided into empirical approach and analytical approaches for the most part as follows:

3.2.1. The Method of Predicting the Maximum Possible displacement of Liquefied Ground Based on Minimum potential energy model

Towhata et al. (1992) proposed a method to predict the maximum possible displacement of liquefied ground using a large-deformation formulation based on the minimum energy principle. Later in 1995 a revised version of a method to predict liquefaction induced subsoil deformation during an earthquake excitation was presented by Towhata et al. (1995&1997). In which, the liquefied soil is treated as a viscous liquid and the deformation is induced by gravity force.

In this method, foundation ground is divided into two layers, one is a non-liquefiable surface layer and the other is the liquefied layer. Firstly the deformation of the liquefied layer is calculated from the state of force balance where the potential energy satisfies the minimum state condition. This deformation is the maximum possible amount of deformation. The vertical displacement of the liquefied layer is assumed to follow a sinusoidal distribution as illustrated in Fig. 3.1.