The purpose of this study is to investigate the mechanism of earthquake-induced failure of dip slopes containing a weak layer by site-specific analysis. Two sites were investigated in this study, including: investigating the properties of materials at the two sites, conducting stability analyses and slope displacement calculations of the two sites.

Background of this study and results from relevant field investigations are the followings:

1) Earthquake-induced landslides are a hazard in many countries in the world, causing billions of dollars of damage and many casualties. The 2004 Niigata-ken Chuetsu Earthquake, with a main shock of Mj=6.8, triggered extensive landslides in Mid Niigata Prefecture, Japan on October 23 in 2004. Two factors are likely to affect the extensive failure of slopes in this earthquake. One is that, before the earthquake, this area had a continuous strong rainfall which was more than 100 mm/day up to 20th October. The other is that, after the mainshock, many large aftershocks struck this area repeatedly. According to aerial photo interpretation, 3,791 slopes failed with total breakdown volume of about 100×106m3 over an area of about 1,310km2. Landslides severely damaged roads, rail lines, life lines, and houses, and dammed streams and rivers.

2) The investigated sites were Yokowatashi site and Higashi-Takezawa site. Yokowatashi site (about 40m wide, 70m long and 4m deep with a slope displacement of over than 50m) and Higashi-Takezawa site (about 295m wide, 350m long and 30m deep with a slope displacement of about 100m) both located near the epicenters of mainshock and aftershocks. Evidence proved that the sliding plane was saturated on both sites during the earthquake. Further more, a weak layer was found on the sliding surfaces of these two failed slopes.

3) Detailed investigation at Yokowatashi site shows that: a) Debris showed that the material on and below sliding plane was within tuff sandy layer which is 1~3 cm thick. Cavities were found in the sandy layer in the vicinity of the failed slope. b) Sliding plane was very planar with a dip angle of about 22°.c) Sandy layer on the sliding Plane had been possibly saturated before the earthquake.

4) Detailed investigation at Higashi-Takezawa site shows that: a) Debris showed the material beneath sliding plane was slightly weathered siltrock. Soft sandy silt was found on a part of sliding surface and in the vicinity of the investigated previous sliding plane. Cavities were not found in the soft sandy silt layer. As most part of the current sliding plane is not exposed, the details of sliding plane need more investigation. b) Sliding plane had a dip angle of 18°at top region and 20°at toe region. c) Ground water level was high during the earthquake.

To investigate the properties of material from Yokowatashi, triaxial compression tests on six groups of undisturbed specimens were performed. Determining the strength of weak layer under saturated conditions is the main purpose of the tests. The details of the test results are described as follows:

1) The strength was likely to be affected by the soil type and degree of weathering. The softrock specimens, which did not contain the sandy layer, had a peak strength of over 3 MPa in unconfined compression tests and TC tests. The specimens with slightly weathered sandy layer had a peak strength of over 500 kPa in untrained triaxial compression tests under an initial effective confining stress of 20 kPa, and the corresponding shear stress ratio mobilized on the sandy layer was over 1.6. The specimens which had been deeply weathered while having a few cavities in the sandy layer, mobilized the peak stress ratio of 1.3 in undrained triaxial compression tests. The specimens which contained deeply weathered sandy layer with many visible cavities, were the weakest among the six groups tested in this present study.

2) Under saturated conditions, the strength parameters mobilized along the weak layer in terms of effective stress were c=0 and φ'=39.0°. As the failed slope has a dip angle of 22°, the slope was stable under normal condition when there was no earthquake and the ground water level was not extremely high. In the cyclic loading tests, possibly affected by the system compliance, liquefaction was not observed. However build-up of excess pore water pressure was observed. The maximum value of excess pore water pressure ratio, which is the ratio of excess pore water pressure to the initial effective vertical stress, is 0.55.

3) The failure patterns were different among the six groups. The specimens without the sandy layer failed like an ordinary softrock. The shear band of specimens having the sandy layer was formed along the boundary between the sandy layer and the lower softrock. For the specimens which contained many visible cavities in the weak layer, the shear band was thicker than the others and crushing of intermediate sandy softrock blocks were observed. The average thickness of the weak layer has a range from about 3 mm to 13 mm.

4) The above failure pattern seems to contribute to the large residual displacement of the slope that was induced by the earthquake. It is reasonable to extrapolate that in the failed slope there were more cavities in the saturated sandy layer. So, when these cavities that were sandwiched between adjacent softrocks with low permeability were compressed during sliding, the pore water pressure could have increased more significantly than the behavior observed in the present laboratory tests.

To investigate the properties of material from Higashi-Takezawa, triaxial compression tests and simple shear tests on undisturbed specimens were performed. Determining the strength of weak layer under saturated conditions is the main purpose of the tests. The details of the test results are described as follows:

1) The specimens from Higashi-Takezawa site had a weak sandy silt layer which was 3 to 32 mm thick according to the measurement after failure. The shear band with an irregular interface formed within this sandy silt layer in all of the triaxial compression tests and in most of the simple shear tests.

2) In triaxial compression tests under saturated conditions, the peak strength mobilized along the weak layer in terms of effective stress was c=0, φ'=36.2° according to triaxial compression test results. As the dip angle was 18°-20°, the slope was stable under normal condition when there was no earthquake and the ground water level was not extremely high. In undrained triaxial compression test, the increment of strain induced by cyclic loading was very limited, and full liquefaction did not occur under undrained cyclic loading conditions.

3) In simple shear tests under saturated conditions, significant strain softening at large deformation was observed in cyclic simple shear tests while keeping the specimen height constant. After accumulation of horizontal displacement Σ|δ| exceeded 60 cm, the value of stress ratio τ(peak)/σ'(vo) became nearly constant at 0.3, which corresponds to φ=16.7°.

To investigate the occurrence of liquefaction within thin weak layer, liquefaction tests on artificial specimens which contain a Toyoura sand layer were performed. The details of the test results are described as follows:

1) For prismatic specimens with a 74 mm-thick sand layer (Dr=60%), during undrained cyclic loading after a drained preshear, liquefaction was observed. When the sand layer was thinner than 32 mm, under the test conditions without minimizing system compliance, no full liquefaction occurred.

2) For prismatic specimens with a 17 mm-thick sand layer (Dr=60%), during undrained cyclic loading without drained preshear, liquefaction was not observed either. Under the test conditions with minimizing system compliance, on the other hand, full liquefaction occurred.

3) For cylindrical specimens, with a 39 mm-thick sand layer (Dr=60%), during undrained cyclic loading after a drained preshear, full liquefaction was observed.

4) Tests results on artificial specimens showed that the occurrence of liquefaction in the thin sand layer was affected by system compliance significantly.

Using the strength parameters determined by triaxial compressions, stability analyses were performed with limit equilibrium methods.

Stability analysis at Yokowatashi site showed that:

1) Stability analysis results by three methods are consistent with each other. Stability analysis methods include Modeling as infinite slope, Modified Janbu Method and Modified Sweden Method.

2) Instability could not occur at normal time without large earthquake and without an extremely high ground water level.

3) After seismic coefficient exceeds 0.3, even if ground water level is on the sandy layer level, the slope will start to move.

4) Even if ground water level was on the sandy layer level, when excess pore water pressure ratio was 0.55 as observed in cyclic undrained triaxial compression tests, this slope can slide with a horizontal seismic coefficient K=0.

Stability analysis at Higashi-Takezawa site showed that:

1) Stability analysis results by three methods show that, safety factor has the same trends. The analysis results showed that ground water level, excess pore water pressure and earthquake force can affect the stability significantly. Possibly due to the over-simplification on curved sliding plane and non-uniform slice height, results yielded by modeling as infinite slope were largely different from the results yielded by the other two methods.

2) Instability does not occur at normal time without large earthquake and without an extremely high ground water level. Without earthquake and with a ground water level 10 m above the sliding plane, the safety factor is above 1.7.

3) After seismic coefficient exceeds 0.35, even if ground water level was on the sandy layer level, the slope will start to move.

4) Even if ground water level was on the previous sliding plane, when excess pore water pressure ratio was 0.6, this slope can slide with a horizontal seismic coefficient K=0. However, such a high value of excess pore water pressure ratio may not be realistic, since the weak layer consisted of sandy silt with a plasticity index of 19.

Using the strength parameters determined by triaxial compressions and/or simple shear tests, displacement calculation were performed with extended Newmark methods where the effects of 1) excess pore water pressure or decreasing of apparent friction angle and 2) the change of sliding plane on yield seismic coefficient Ky were considered.

Results of displacement calculation of Yokowatashi site showed that:

1) Three methods result in similar displacements under the condition that excess pore water pressure ratio equals zero. The methods include Modeling as infinite slope, Newmark with Janbu Method and Newmark with Sweden Method.

2) Excess pore water pressure has extremely significant effect on displacement. The calculation result is similar to the measured value by using an excess pore water pressure ratio of 0.55. However, the displacement by numerical calculation is 0.9~2.1 m if excess pore water pressure ratio equals zero.

3) Ground water level has significant effect on displacement. If ground water level is 2 m above the sliding plane, the displacement is about 100 m with an excess pore water pressure ratio of 0.55 by Newmark with Janbu Method. However, it is about 70 m if ground water level is on the sliding plane.

Results of displacement calculation of Higashi-Takezawa site showed that:

1) Modified Newmark with Sweden Method and Modified Newmark with Janbu Method result in similar displacements when the calculated displacements are small. However the difference between the calculated displacements by these two methods is large when the calculated displacement is more than 25 m. A possible reason is the effects of shape of sliding plane.

2) Excess pore water pressure has extremely significant effect on displacement. By using the decreasing friction angle obtained from cyclic simple shear tests, the calculation results in a displacement of about 59 m by Modified Newmark with Janbu Method. However, if the friction angle is kept constant as its peak value, the displacement is about 0.5 m.

3) Ground water level has significant effect on displacement. If ground water level is 12 m above the sliding plane, the residual displacement can be about 59 m by Modified Newmark with Janbu Method. However, it will be only 4.6 m if ground water level is on the sliding plane.

According to the displacement calculation results and stability analysis results, with an excess pore water pressure ratio of 0.55 and an ground water level about 1 m above the sliding plane, the failure mechanism of Yokowatashi site can be inferred as follows:

At normal time without large earthquake and without an extremely high ground water level, since the silt layer on the previous sliding plane has a friction angle of 39.0°, the slope is stable.

From three days before the 2004 Niigata-ken Chuetsu Earthquake, the excessive antecedent rainfalls saturated the sandy layer and raised the ground water level. The safety factor was decreased by this precipitation. During the earthquake the excess pore water pressure increased significantly, the slope started to slide along the sandy layer with a low yield seismic coefficient Ky. The maximum velocity by simulation was about 10 m/s. After the sliding mass reached the road location at the toe of the slope, Ky began to increase rapidly. Almost simultaneously seismic motion ceased. After having horizontally slid about 60 m, the sliding stopped.

The key point for the long residual displacement of this slope is the build-up of excess pore water pressure.

According to the displacement calculation results and stability analysis results, the failure mechanism of Higashi-Takezawa site can be inferred as follows:

At normal time without large earthquake and without an extremely high ground water level, since the silt layer on the previous sliding plane has a friction angle of 36.2°, the slope is stable.

From three days before the 2004 Niigata-ken Chuetsu Earthquake, the excessive antecedent rainfalls saturated the weak silt layer and raised the ground water level. The safety factor was decreased by this precipitation. During the earthquake, the slope started to slide mainly along the saturated weak layer. Since the friction angle decreased with the slope displacement, the yield seismic coefficient Ky decreased to a value below zero. This negative Ky accelerated the sliding. The maximum velocity by simulation was about 7 m/s. After the sliding mass reached the riverbed at the toe of the slope, Ky began to increase rapidly. Almost simultaneously seismic motion decreased from this time. After having horizontally slid about 110 m, the movement of slope was obstructed by the slope and embankment on the other side.

The key point for the long residual displacement of this slope is the decreasing friction angle.

本論文はCase studies on the mechanism of earthquake-induced failure of dip slopes containing a weak layer(弱層を含む流れ盤斜面の地震時崩壊メカニズムに関する事例研究)と題した英文の論文である。

2004年の新潟県中越地震では、斜面崩壊が多数生じて河道閉塞や家屋・道路等の被害を引き起こした。被災した地域は活褶曲地帯であったために、斜面と平行な地層構造を有する流れ盤構造が崩壊した事例も多かった。特に、小千谷市横渡と旧山古志村東竹沢で発生した斜面崩壊は、流れ盤構造であったことに加えて、地層中の既存の弱層をすべり面とした崩壊が生じた点が特徴的であった。

これらのような弱層を含む流れ盤斜面の地震時崩壊現象については、地盤工学的な観点から検討を行った例は限定的であり、弱層の強度特性等については未解明な点が多い。また、すべり破壊が生じた場合にどの程度まですべり変位が発生するかについての検討は十分には行われていない。特に、2004年新潟県中越地震では、地震発生の数目前に大量の降雨があり、その影響を考慮した検討を実施する必要がある。

以上の背景のもとで、本研究では前述した横渡と東竹沢の2箇所における斜面崩壊を対象として、原位置で採取した乱れの少ない試料を用いた室内土質試験を系統的に行って弱層の強度特性を明らかにするとともに、その結果を用いた安定解析と地震時変位解析を行い、最終的にはこれらの斜面の地震時崩壊メカニズムを解明することを目的とした事例検討を実施している。

第一章では、既往の研究を整理したうえで本研究の目的を設定し、論文全体の構成について説明している。

第二章では、関連研究として別途実施された原位置調査の結果のうち、本研究の内容に関わる成果についてとりまとめている。

第三章では、試験装置と試験方法および試験に用いた試料の詳細について記述している。各斜面において今回の地震ではすべり破壊の生じなかった位置で弱層を含む試料をブロックサンプリングにより採取し、主として三軸試験に供している。さらに、横渡ではボーリング孔を利用した試料採取も行って、ブロックサンプリング試料との比較を行っている。一方で、東竹沢で採取したブロックサンプリング試料については、その一部を用いて単純せん断試験も実施している。

第四章では、試験結果を提示してその考察を行っている。単調載荷試験の結果に基づいて強度定数を求めるとともに、非排水繰り返し載荷試験の結果に基づいて地震中の過剰間隙水圧の発生量を評価している。破壊時には弱層にひずみの局所化が生じることを示し、弱層面の角度が供試体ごとにばらつく点を考慮してせん断抵抗角の評価を行っている。横渡の弱層は一部空洞を含む凝灰岩質のゆるい砂層であるために、地震中に過剰間隙水圧が初期有効応力の55%程度まで発生する可能性があることを示す一方で、東竹沢の弱層は塑性指数19の砂質シルトであるために地震中の過剰間隙水圧の発生量は限定的であることを示している。東竹沢の試料については繰り返し単純せん断試験も実施して、累積せん断変位の増加とともに明確なひずみ軟化現象が生じ、その特性は繰り返しせん断変位振幅や高拘束圧の違いによらないことを明らかにしている。さらに、厚さの異なる砂層を有する人工供試体の繰り返し三軸試験も実施して、試験結果に及ぼすシステムコンプライアンスの影響を定量的に評価している。

第五章では、試験結果として得られた弱層の強度特性を用いた安定解析の結果を記述している。地震前には降雨により地下水位が上昇しても斜面の安定が保たれる一方で、地震時には水平震度0.3~0.35相当の地震力が作用することにより斜面がすべり破壊をおこすという結果が得られ、実際の挙動と整合することを見出している。

第六章では、すべり破壊が生じた後の地震時変位量の解析結果について記述している。横渡の事例に対しては、飽和した弱層で過剰間隙水圧が発生する影響を考慮し、さらに地形の影響も考慮することにより実事例と同程度な大変位が解析でも得られることを示している。また、東竹沢の事例に対しては、弱層のひずみ軟化特性を考慮し、さらに地形の影響も考慮することにより実事例よりやや小さい変位が解析で得られることを明らかにしている。東竹沢の事例で解析値が過小評価された理由として、試料を採取した位置での弱層の特性が、必ずしもすべり面全体の特性を代表していない可能性を指摘している。

第七章では、本研究で得られた成果を結論としてまとめ、今後の課題を整理している。

以上をまとめると、本研究では、弱層を含む流れ盤斜面が崩壊した2事例を対象として、室内土質試験により弱層の強度特性を測定するとともに、安定解析と地震時変位解析により地震前の降雨の影響と地震動自体の影響を明らかにし、これらの結果に基づいて地震時崩壊メカニズムを解明している。このことは地盤工学の進歩への重要な貢献である。よって本論文は博士(工学)の学位請求論文として合格と認められる。