EARTHQUAKE-INDUCED SUBMARINE LANDSLIDES IN VIEW OF VOID REDISTRIBUTION

Transcription

1 Liu, Deng and Chu (eds) 2008 Science Press Beijing and Springer-Verlag GmbH Berlin Heidelberg Geotechnical Engineering for Disaster Mitigation and Rehabilitation EARTHQUAKE-INDUCED SUBMARINE LANDSLIDES IN VIEW OF VOID REDISTRIBUTION Takaji Kokusho Faculty of Science & Engineering, Chuo University, Kasuga, Bunkyoku Tokyo, Japan Tetsuya Takahashi Ex-Graduate Student, ditto Submarine slides occurring near-shore in non-cohesive sandy deposits are first discussed in view of void redistribution effect. Model tests demonstrates that formation of water film beneath silt seam plays a key role in liquefaction-induced sliding failure in gentle slope. Considering that soil deposits are naturally stratified with sandwiched low permeability seams, it seems quite reasonable to recognize the water film effect as a major mechanism for seismically induced submarine slides in gently sloped sandy sea-bed near coasts. Analogous void redistribution effect on the submarine slides occurring in clayey sea floor far from coasts is then studied by cyclic loading tests to find out large volumetric strain after strong cyclic loading. Simple consolidation plus seepage analysis considering the earthquake-induced large volumetric strain indicates that excess pore water may accumulate beneath less permeable seams sandwiched in relatively permeable clayey layer, possibly causing delayed shear strength reduction due to excessive swelling just beneath the seam. INTRODUCTION One of geotechnical hazards on which geotechnical engineering has not focused so much is submarine slides. Various causes are considered as the triggers; earthquakes, high ocean waves during storms, tidal changes, etc., among which earthquake-induced slides are focused here. The slides may be classified into two types. The first is the ones which occur near coasts, affecting coastal areas in deltaic lands or man-made fills. The slopes in those slides are relatively steep compared to the other type. For example, Valdez and Seward, port cities in Alaska, USA, suffered great loss of human lives and properties by large scale submarine slides involving coastal areas (Coulter et al., 1966; Lemke et al., 1967). As shown in Figure 1, the inclination of the sea bed was 30 at most near the beach, but actually 5 or less on average considering the long span of the slip surface of more than 2 km offshore, which is considerably less than the internal friction angle of the soil. Analogous submarine failure occurred during the 1999 Kocaeli earthquake along the southern coast of the Izmit bay in Turkey (The Japanese Geotechnical Society, 1999). The slope of the seabed originally

2 about 5-10 consisting of coarse sand and gravel became very steep at the scarp after the earthquake as shown in Figure 2. In the world, there remain quite a few coastal zones with high seismicity and relatively steep sea floor consisting of loose sand or gravel where submarine slide is likely to occur. Figure 1. Cross-sectional change of sea bed at Valdezbefore and after the 1964 Alaskan earthquake (Coulter 1966) Figure 2. Cross-sectional change of sea bed at Valdez before and after the 1999 Kocaeri earthquake (JGS 2000) The second type is the submarine slides which occur offshore in the deep sea bottom. The slope is normally gentler than a few percent or even less than 1% (Hampton and Lee, 1996). A typical example of this type occurred 60 km off the California Coast during a 1980 medium magnitude earthquake (Field et al., 1982). In this case, 2 by 20 km sea floor consisting of sand and mud of only 0.25 degrees slipped and became much more flat with some evidence of liquefaction such as sand boils on the sea bottom of 30 to 70 m depth as sketched in Figure 3. Figure 3. Cross-sectional change of sea bed off the California Coast before and after a 1980 medium magnitude earthquake (Field et al., 1982) 178

3 As one of possible mechanisms to explain sliding failures of low-gradient sandy slopes, void redistribution effect may be considered. In this view, fine soil sublayers sandwiched in sand deposits are considered to play a key role in flow failure. The formation of water interlayers (Seed, 1987) or water films between liquefied sand and overlying lower-permeability seams has been observed under level ground conditions in a number of model tests. For sloping ground conditions it has been demonstrated, based on model tests, that the water film or void-redistribution effect plays an important role in post-earthquake large lateral flow in liquefied ground. Thus, liquefaction may be highly responsible in earthquake-induced submarine slides, particularly in near-shore sites where the seabed is composed of liquefiable loose sand or gravel. However, in off-shore slides far from coasts, sea floors are normally very gently sloped and composed of cohesive fine soils, in which liquefaction seems to hardly occur. Nevertheless, slumps and scars of off-shore submarine slides are often found and sometimes really huge in scale and suspected as potential causes of tsunami other than fault dislocations. Thus, a concern on submarine slides from geotechnical point of view is necessary as one of potential natural disasters for coastal structures, offshore explorations for natural resource and even for tsunami. Needless to say, much more research is needed to understand the mechanism correctly. In the first part of this article, the void redistribution effect in liquefied sand slopes will be addressed as one of the key mechanism of submarine slides particularly near coastal areas where seabed is mostly sandy and non-cohesive. Then, off-shore conditions far from coastal areas will be studied to examine a relevance of a similar void redistribution mechanism in cohesive soils for major cause of submarine slides. VOID REDISTRIBUTION MECHANISM IN LIQUEFIED SAND SLOPE Kokusho (1999) demonstrated, by model tests, soil element tests and site investigations, that water film or void-redistribution effect plays an important role in post-earthquake large lateral flow in liquefied ground. This concept had been introduced in a committee report in US (National Research Council, 1985) and also discussed by Seed (1987) by using a special term, water interlayer. In this view, fine soil sublayers sandwiched in sand deposits are considered to play a key role in flow failure. A sand layer, which is classified as a single uniform layer in normal engineering practice, may also be composed of sublayers with different grain size and, hence, different permeability. Thus, soil stratification may be able to Figure 4. Shake table test for 2D saturated sand slope sandwiching silt arc 179

4 explain, as Seed (1987) already pointed out, why steady-state strength of uniform sand leads significantly higher values of residual strength than those estimated by case studies in the field. Liquefaction-induced lateral flow in gentle slopes have been studied by the present author and his research group (Kokusho, 1999, 2000, 2003, Kokusho and Kojima, 2002; Kokusho and Fujita, 2002). focusing on the significant effect of void redistribution or water film by utilizing various investigation methods such as soil investigations of insitu deposits, 1-D liquefaction tests in a tube, 1G shake table tests, laboratory soil tests, field survey, numerical analyses etc. Among them, 2-dimensional shaking table tests were performed as shown in Figure 4. Clean fine sand was rained in water to make saturated loose sand slope in a rectangular soil box in which an arc-shape silt seam was sandwiched. The model was subjected to 3 cycles of sinusoidal shaking perpendicular or parallel to the sloping direction and the sliding was observed through the transparent side wall. It was demonstrated that soil mass flows along a continuously formed water film beneath the arc-shaped fine soil seam still after the end of shaking, while in a uniform sand without a silt seam a major slide takes place only during shaking. Video movies of the tests are accessible at the web site; http//: Figure 5(a) indicates that the slide occurs not only during shaking but also after the end of shaking quite discontinuously along the silt arc beneath which thin water film could be recognized. In contrast, Figure 5(b) indicates that flow deformation in a uniform sand Figure 5. Cross-sectional deformation for slopes with silt arc (a) and without silt arc (b) Top of (a) and (b): during shaking, Bottom of (a) and (b): after shaking model occurs mostly during shaking continuously in the cross-section. Time histories of the same test results are shown in Figure 6. In Figure 6(a) without the silt seam, flow is 180

5 restricted during shaking, while in Figure 6(b) with the silt arc large flow also occurs after shaking. The target points on the charts are shown in Figure 6(d) by the same symbols. These results are for the input acceleration of 0.31 G. In Figure 6(c), the time histories of the flow deformation of the same model subjected to weaker input acceleration of 0.18G. Interestingly enough, much larger post-shaking flow occurs than the case of the larger input acceleration of 0.34, and only minimal deformation takes place during shaking. This is because in the weaker motion, the slope remains steep leaving larger driving force for the post-shaking flow along the water film, if it is formed. Figure 6. Time-dependent sliding displacement at target points shown in (d); without silt arc by Acc G (a), with silt arc by Acc 0.34 G (b) and with silt arc by Acc G(c) A basic question may arise that sand which can be so dilative if sheared under a low confining stress may absorb ambient excess pore water and hence block the water film development. It was pointed out, however, based on the comparative observation of the cases with and without a silt seam that a water film formed beneath the seam serves as a shear stress isolator which shields the deeper soil from the development of shear strain and dilatancy (Kokusho, 2000). Consequently, sand can experience large shear strain beneath the silt seam without suffering from the dilatancy effect, whereas it stops moving after the end of shaking if the sand is uniform. In another shaking table test, a soil mass slid even on a very gently inclined water film, which broke at weak points of the overlying sublayer, triggering the boiling failure in the sand above and a mud avalanche of the upper layer (Kokusho, 1999; Kokusho, 2000). A video movie of the test is also accessible at the web site. Considering that no such drastic sliding failure occurs in the uniform sand slope, a significant effect of the void 181

6 redistribution or water film formation has been demonstrated. Considering that soil deposits are naturally stratified with sandwiched low permeability seams, it seems quite reasonable to identify the water film effect as a major mechanism for seismically induced submarine slides in gently sloped sandy seafloor near coast. If water films are formed continuously, they will tremendously reduce the residual strength. If sliding occurs all the way through a continuous water film, the strength ratio becomes zero. This however seems unrealistic because water films may be neither continuous nor straight but winding with rough faces. Based on model shaking table tests, Kabasawa and Kokusho (2003) quantified the residual strength exerted during the delayed flow along a water film by using the balance between potential energy and dissipated energy along the slip surface. The result shown in Figure 7 indicates that Figure 7. Residual strength along water film evaluated from model shake table tests with different conditions (Kabasawa & Kokusho, 2005) the residual strength along the water film is almost independent of sand density and other test parameters and remains around 20% that of the uniform sand. POST-EARTHQUAKE VOID REDISTRIBUTION IN COHESIVE SEABED Figure 8(a) and (b) show relationships of water depth versus water content (a) and water content versus clay content (b) for submarine soils in Japan Sea (Ikehara, 1989). By combining the two charts, a clear trend can be recognized that the soil is changing from sandy silt to clay with increasing water depth or increasing distance from coast. Geotechnical properties of top sediments in Korea Plateau margin in Japan Sea are exemplified in Table 1 (Lee et al., 1991), which clearly indicates again that soils are changing from sandy silt to clay with increasing water depth from near-shore continental shelf (<200 m) to offshore continental slope (>1500 m). Water content and plastic index tend to increase accordingly toward the offshore direction, being consistent with the change in soil type. Quite a few slump scars can be found in lower continental slope in Japan Sea where the slope inclination is 1-2. This indicates that multiple massive submarine landslides had occurred from time to time presumably by earthquakes (Lee et al., 1991). Thus, it is not exceptional that submarine slides occur not only near the coast but also at continental slopes of deep sea where the seabed deposits are predominantly clayey soils. 182

7 Table 1. Physical properties of topmost sediments in Japan Sea (Lee et al., 1991) Province Solicontent(%) Sand Silt Clay Water content W(%) Plasticity index Ip Sensitivity ratio Shelf ( 200m) Upper slope ( m) Middle slope ( m) Lower slope ( 1500m) (a) (b) Figure 8. Relationships of water depth versus water content (a) and water content versus clay content (b) measured for seabed soils in Japan Sea (Ikehara, 1989) In order to investigate the cyclic loading effect during strong earthquakes on submarine soft clay, undrained cyclic triaxial tests were performed for intact soil samples recovered from Japan sea off West Japan. The basic properties of tested specimens are listed in Table 2. The depth of the sampled soils are from 60 to 100 cm from the sea floor. Judging from the water content of the sample, w=73%-90%, it is estimated from previous research results (Ikehara, 1989) that the soil is from continental slope. The initial void ratio is e= and the plasticity index is I P = The test specimen of 50 mm in diameter and 100 mm in height was set in the triaxial apparatus shown in Figure 9. It was first consolidated by the confining stress σ c =29 kpa corresponding to the shallow sampling depth and then loaded in 30 cycles under undrained condition by cyclic axial stress σ d with frequency of 0.05 Hz. The stress ratio (R= σ 2σ ) was varied from 0.23 to d c In Figure 10(a), pore-pressure ratio u σ c at the end of cyclic loading is plotted against the stress ratio R. The pore Figure 9. Triaxial apparatus used for cyclic loading tests of submarine clay 183

8 pressure builds up to u σ c 90% for the stress ratio around R 0.5 and then approaches to an asymptote. After the cyclic loading of 30 cycles, the specimen was drained and the volumetric strain ε v was measured by excessive pore water. Figure 10(b) shows the plots of volumetric strain ε v against stress ratio R. It is noted that if the stress ratio attains R= , the post-earthquake volumetric strain reaches ε v =4%-5%. This magnitude of volumetric strain is almost equivalent to that of loose sand after liquefaction, which actually occurred during the 1964 Niigata earthquake or the 1995 Kobe earthquake, indicating that the effect of the excess pore-water may not be ignored in view of the void redistribution even in cohesive submarine clay. Table 2. Physical properties of specimens for cyclic triaxial tests Depth(cm) Void ratio before test e Water content W(%) Particle density ρ (t/m 3 ) s LL(%) PL(%) PI(%) Figure 10. Stress ratio plotted versus pore pressure ratio (a) and volumetric strain (b) for submarine soft clay by cyclic triaxial tests Analogous to sandy deposits in which the soil is normally stratified as demonstrated by Kokusho (2003), it may well be estimated that submarine clay layer is also strongly stratified. Sonic profiles of submarine clay layer normally show a set of reflection boundaries indicating such stratification. It further implies that clay layer similar to sand layers consist of sublayers with different properties and different permeability. In order to know the role of the excess pore-water seismically squeezed in stratified clayey submarine deposits, a simple model shown in Figure 11, which is basically the same as in the previous research for analyzing water film effect in liquefied sand (Kokusho 2000), was chosen for consolidation analyses. The model consists of upper and lower clay layers of low permeability sandwiching a seam of still lower permeability in between.

9 Linear consolidation by self weight is assumed to occur in parallel in the two layers from the initial excess pore-pressure of triangular shape corresponding to 100% pore pressure buildup due to cyclic loading as shown in Figure 11. In the process of consolidation, excess pore-water in the lower layer migrates upward and accumulates beneath the seam developing a narrow band of higher water content. The pore-pressure there is sustained equal to the effective overburden, while the pore pressure at the bottom in the upper layer continues to decrease with the progress of consolidation, introducing seepage flow through the intermediate seam. Even after the completion of the consolidation in the lower layer, the seepage flow continues until all the accumulated pore-water beneath the seam goes out to the surface. In the analysis, the consolidations in the individual layers are computed in parallel assuming impermeable lower boundary and then seepage velocity is calculated from hydraulic gradient in the seam. The seepage velocity and pressure are superposed together to evaluate the accumulation of excess pore-water beneath the seam during the consolidation and seepage process. Figure 11. Analytical model for void redistribution in clay deposit sandwiching seam of lower permeability Eight cases, Case1-8, were analyzed, the parameters of which are listed in Table 3. The thickness of intermediate seam was assumed 0.2 m in most cases but also changed as 0.2 m or 2.0 m in Case 7 and Case 8, respectively. Permeability in the upper/lower layer and the seam was systematically changed in the range of cm/s as listed in the table based on consolidation tests conducted on the same soil (Takahashi 2007), so that the seam always has a lower value than the sandwiching upper and lower layers by the ratio of 10-1 ~10-3. In all the cases, the thickness of the upper and lower layers were set as 10 m and the post-earthquake volumetric strain was prescribed as 5% based on the test results shown in Figure 10(b) corresponding to large cyclic stress ratio of R=0.5. Consequently, the compressibility coefficient was evaluated as m v = kpa -1. The buoyant unit weight of the clay was assumed as 4.0 kn/m

10 Table 3. Analytical cases for clay seafloor sandwiching more impermeable silt seam considering void redistribution effect Upper layer Intemediate seam Lowerlayer Case Pemeability Thickness(m) (cm/s) Thickness(m) Pemeability(cm/s) Thickness(m) Pemeability(cm/s) Figure 12 shows major results of the analysis; time-dependent variation of excess pore-water accumulation beneath the intermediate seam expressed in terms of equivalent pure water thickness. It is obvious that no meaningful water accumulation occurs in Cases 1, 4, 6 where the permeability ratio between the seam and the sandwiching layers is 10-1, while Cases 2, 3, 5 indicate that the permeability ratio lower than 10-2 leads to water accumulation equivalent to or more than 10 cm water thickness and it becomes thicker for the permeability ratio getting smaller from 10-2 to It should be noted that the pore-water accumulation sustains very long, 150 days (Case 2), 400 days (Case 3) and 1600 days (Case 5) if the total soil system becomes more impermeable at the same time keeping their ratios constant. It is also obvious by comparing Case 2, 7, 8 that the duration of the water accumulation very much reflects the thickness of intermediate seam. Figure 12. Time-dependent variation of excess pore-water accumulation beneath the intermediate seam expressed in terms of equivalent pure water thickness Thus, the simple analyses on the sea-bottom clay layers sandwiching lower permeability seam indicate the occurrence of void redistribution which may be equivalent to more than 10 cm water thickness and sustain for hundreds of days in the condition given here. The thickness and time duration may vary according to various conditions such as seismic stress 186

11 ratio, thickness and compressibility of the seabed soft clay, etc. However, the effect seems great enough to affect the post-earthquake stability of submarine gentle slope. A big question here is how the excess pore-water will be accumulated beneath the seam. If the soil is non-cohesive, it is clear that a thin water-interlayer will be formed causing sliding failure along it as demonstrated by Kokusho (1999). If the soil is highly plastic and cohesive as observed in many field research, the cohesion will prevent the formation of water film and swelling of clay will occur. As the quantity of the accumulated pore water gets larger, the soil will excessively swell leading to very large void ratio and very small shear strength. More research is certainly needed to clarify what will happen in submarine clays and how the strength is affected beneath a very low permeable seam if excessive pore-water is supplied from below. CONCLUSIONS In the first part, the mechanism of submarine slides occurring near-shore in cohesionless sandy deposits was discussed. Model tests demonstrated that the void redistribution or water film effect plays a key role in sliding failure in liquefied gentle slope. The residual strength along the water film evaluated in the model was found very much reduced down to 20% of the original strength due to the effect. Considering that soil deposits are naturally stratified with sandwiched low permeability seams, it seems quite reasonable to identify the water film effect as a major mechanism for seismically induced submarine slides in gently sloped sandy or gravelly sea-bed near coastal areas. Analogous void redistribution effect on the submarine slides occurring in clayey sea floor was investigated on samples from Japan Sea to find out that the volumetric strain after strong cyclic loading was almost equivalent to that in liquefiable loose sand. A simple analysis combining consolidation and seepage flow revealed that large quantity of excess pore water accumulates beneath a more impermeable seam sandwiched in clay deposits, causing swelling of clay just beneath the seam. More research is needed whether or not the excessive swelling gives significant effect on delayed instability of submarine slope consisting of plastic soft clay. ACKNOWLEDGMENTS Dr. Ken Ikehara of Geological Survey and Applied Geo-Science, AIST, Tsukuba, Japan, who kindly provided submarine soil samples from Japan Sea and also valuable advice on marine geology is gratefully acknowledged. REFERENCES Coulter H. W., Migliaccio R. R. (1966). Effects of the Earthquake of March 27, 1964 at Valdez, Alaska. Geological Survey Professional Paper 542-C, U. S. Department of the Interior, p.36. Field M. E., Gardner J. V., Jennings A. E. and Edwards B. D. (1982). Earthquake-induced sediment failures on a 0.25 slope, Klamath River delta, California. Geology, 10: Hampton M. A., Lee H. J. (1996). Submarine landslides. Reviews of Geophysics, 34:

12 Ikehara K. (1989). Some physical properties of shelf to basin deposits off San in and Hokuriku district, southern part of Japan Sea. Bulletin of Geological Survey of Japan. 40 (5): (in Japanese). Kabasawa K. and Kokusho T. (2003). Energy analysis and model tests on lateral flow induced by water film effect in liquefied ground. Journal of Japan Society for Civil Engineers, N0.771/III-68, (in Japanese). Kokusho T. (1999). Formation of water film in liquefied sand and its effect on lateral spread. Journal of Geotechnical and Geoenvironmental Engineering, ASCE, 125(10): Kokusho T. (2000). Mechanism for water film generation and lateral flow in liquefied sand layer. Soils and Foundations, 40(5): pp Kokusho T., Fujita K. (2002). Site investigation for involvement of water films in lateral flow in liquefied ground. Journal of Geotechnical and Geoenvironmental Engineering, ASCE, 125(10): Kokusho T., Kojima T. (2002). Mechanism for post-liquefaction water film generation in layered sand. Journal of Geotechnical and Geoenvironmental Engineering, ASCE, 128(2): Kokusho T. (2003). Current state of research on flow failure considering void redistribution in liquefied deposits. Soil Dynamics and Earthquake Engineering, Elsevier, 23: Lee H. J., Clough S. S., Chun S. S. and Han S. J. (1991). Sediment failures on the Korea Plateau slope, East Sea (Sea of Japan). Marine Geology, 97: Lemke R. W. (1967). Effects of the Earthquake of March 27, 1964 at Seward, Alaska. Geological Survey Professional Paper 542-E, US Department of Interior. National Research Council. (1985). Liquefaction of soils during earthquakes. Committee of Earthquake Engineering, Commission of Engineering and Technical Systems, National Academy Press, Washington, D.C. Seed H.B. (1987). Design problems in soil liquefaction. Journal of Geotechnical Engineering, 113(8): ASCE. Takahashi T. (2007). Geotechnical properties of submarine sediments and basic study on submarine landslide mechanism. Master s Thesis, Graduate School of Chuo University (in Japanese). The Japanese Geotechnical Society. (1999). Investigations of the 1999 Kocaeli earthquake. Reports on the Investigations of the 1999 Kocaeli Earthquake in Turkey and the 1999 Chi Chi Earthquake in Taiwan, JGS (in Japanese). 188