ICSE6-158 A Case Study on Seepage Failure of Bottom Soil within a Double- Sheet-Pile-Wall-Type Ditch Tsutomu TANAKA 1, Wataru TAKASHIMA 2, Tran Thi Hanh PHAM 1, Ken URATA 3 and Nobuhiro UEMURA 4 1 Department of Agricultural and Environmental Engineering, Kobe University Nada Kobe 657-8501, Japan: ttanaka@kobe-u.ac.jp : 2 Kamo Region Agriculture and Forestry Office, Gifu Prefecture Furui Minokamo Gifu 505-8508, Japan: takashima-wataru@pref.gifu.lg.jp : 3 Disaster Prevention and Restoration Division, Rural Development Bureau, MAFF Kasumigaseki Chiyoda-ku Tokyo 100-8950, Japan: ken_urata@nm.maff.go.jp : 4 River and Harbor Section, City Development and Planning Bureau, Amagasaki City Higashi-Nanahonmatsu Amagasaki 660-8501, Japan: uemura-nobuhiro@city.amagasaki.hyogo.jp : Abstract: Boiling occurred in the bottom soil within a double sheet-pile-wall-type ditch (or prefabricated ditch) during puddling, and the ditch border collapsed. In this case, it was thought the penetration depth was so limited that the bottom soil of the ditch became unstable when the groundwater level increased. In this paper, FEM seepage flow and stability analyses were conducted, and the safety factors against seepage failure of ditch bottom soil were calculated. Causes of the failure of the ditch border, effects of two-dimensionally concentrated flow, effects of backfills, and countermeasures against seepage failure were discussed. Key words Prefabricated ditch, Seepage failure, FEM seepage flow and stability analyses, Backfills, Countermeasures. I INTRODUCTION Boiling occurred in the bottom soil within a double sheet-pile-wall-type ditch (or prefabricated ditch) during puddling, and the ditch border collapsed as shown in Figure 1. The ditch is located at the west part of the Nobi Plain which was developed as a backmarsh or alluvial lowland of Nagara River and Kiso River in Gifu Prefecture, Japan. The surface soil consists of alternating fine sand and silt layers. In this case, the penetration depth was so limited that the bottom soil of the ditch became unstable when the groundwater level increased. Changes in the levels of groundwater and water level in the ditch were observed continuously from 2003/12/05 to 2004/04/03. Intermittent observations were subsequently made at several events such as puddling, transplanting, heavy rain, etc. In agricultural land in Japan, puddling (i.e. ploughing and irrigating the fields) is conducted before transplanting rice seedlings. In this paper, the two hydraulic head differences, H, were considered: (1) 0.212m on 2003/12/12 as the largest value during the continuous observation period, and (2) 0.456m on 2004/05/22 during puddling. The greatest H value at the time of transplanting was estimated to be 0.878m on 2004/05/24, when the boiling failure occurred. Figure 1: Collapse of ditch border For the above two conditions, FEM seepage flow and (2001/06/19, during transplanting) stability analyses were conducted, and the safety factors 1543
against seepage failure were calculated. Causes of the failure of the ditch border, effects of two dimensionally concentrated flow, effects of backfills, and countermeasures against seepage failure were discussed. II DETAILS OF GEOLOGY AND DITCH CONSTRUCTION Figure 2 shows the site of the ditch where boiling occurred. Piping failure occurred in the bottom soil of the ditch near boring Bo.1, but did not occur near boring Bo.2 [Tanaka et al. 2011b] which is located 240m from Bo.1. Figure 3 shows a cross-sectional view of the ditch near Bo.1. The horizontal soil layers are, from top downward: plow layer, fine sand containing silt, silt and fine sand containing silt. The N-values of these soils are very low. In an agricultural field improvement project started in 1997, the farm lands were re-divided using 1 ha (50m 200m) as a unit. To secure the biodiversity of the ditch with conservation and improvement of the habitat, the bottom soil was left in the ditch without concrete slab or lining. The double-sheet-pile-wall-type ditch (or prefabricated ditch) was constructed. The pre- Figure 2: Plane of the site fabricated ditch consists of two parts: a panel and arm. The width of the arm (1,000mm at the topmost level) is narrower than that of the panel (1,140mm at the same level). The wall of the ditch is 1.2m high, and is composed of 4 panels (each panel has a height of 0.3m). III PYSICAL PROPERTIES OF SOILS Table 1 shows the physical properties of soils obtained by boring Bo.1, backfill and boiled-out sand at the ditch bottom, and Figure 4 shows the grain size distributions of soils. It is found from Table 1 and Figure 4 that the backfill is classified as the same soil as the upper fine sand containing silt. In boring exploration for the backfill, a drop hammer sank under its own weight and the N-value was evaluated as 0. The backfill is very soft, and is also more permeable than neighboring soils in the site. The boiled-out sand is thought to be the upper silt containing sand whose fine particles are washed out by flowing water in the ditch. Figure 3: Cross sectional view of the ditch near Bo.1 1544
Table 1: Physical properties of soils Plow layer EL.3.20~2.40 EL.2.40~-0.40 EL.-0.40~-1.95 Backfills Boiled-out sand (Boring log) - (Fine sand with silt) (Silt) (Fine sand with silt) - - G s 2.641 2.657 2.656 2.658 2.655 2.651 U c 29.196 15.271 14.377 13.963 13.973 2.544 D 50 (mm) 0.0424 0.172 0.00770 0.0598 0.175 0.227 e max - 1.290-1.585 1.242 1.231 e min - 0.521-0.701 0.616 0.670 w L 25.215-41.209 - - - w P 22.140-29.032 - - - k 15 (cm/s) 2.712 10-5 1.310 10-3 7.659 10-6 1.466 10-4 2.265 10-3 4.054 10-3 Classification Silt Silty sand Silt Silt Silty sand Poorly graded sand Percent finer by weight (%) 100 80 60 40 20 Plow layer Upper fine sand containing silt Silt Lower fine sand containig silt Back fill Boiled-out sand 0 0.001 0.010 0.100 1.000 Grain size (mm) Figure 4: Grain size distributions of soils IV OBSERVATION OF GROUNDWATER LEVEL AND WATER LEVEL IN THE DITCH AT BORING BO.1 The groundwater levels and water levels in the ditch were continuously observed during 2003/12/05 to 2004/04/03. From observations of the heights of water levels and precipitation at Gifu Weather Station, it can be seen that the ground water level tends to become higher after a rainfall, and subsequently the head difference in water levels between up- and downstream sides, H, becomes high. The largest head difference H is recorded as 0.212m on 2003/12/11 during the continuous observation period. Intermittent observation was carried out after the continuous observation period. Surface boiling was observed on the ditch bottom as shown in Figure 5, when the ground water Figure 5: Surface boiling (2003/08/24) 1545
level was high, e.g. at H = 0.317m on 2004/11/20, H = 0.373m on 2005/11/12, H = 0.273m on 2006/04/23. H = 0.600m was recorded on 2004/05/16 after heavy rain, and H = 0.456m on 2004/05/22 during puddling. Two days after the latter, on 2004/05/24, the ditch border again collapsed about 10m from the observation point Bo.1 as shown in Figure 6. The head difference H at the time of failure was estimated to be 0.878m. V SEEPAGE FLOW ANALYSES Cases of the two hydraulic head differences, H, were analyzed: (1) 0.212m on 2003/12/12 as the largest value during the continuous observation period, and (2) 0.456m on 2004/05/22 during puddling. Let us consider the first case of H = 0.212m on 2003/12/12 as shown in Figures 7 (a) and (b) when not considering and considering backfills, respectively. Flow nets are given by Figures 8 (a) and (b) when not considering and considering backfills, respectively. A finite element mesh with 5,756 nodes and 5,752 elements is used in the analysis, and equi-potentials are drawn at an increment of one tenth of the head difference H. Figure 8 (b) shows flow nets for the case that the backfills were more permeable than neighboring soils (k BF = 2.265 10 5 m/s (real)). Figure 8 (c) shows flow nets for an assumed condition that the backfills have about one order smaller permeability (k BF = 1.31 10 6 m/s (assumed)). Figure 8 (b) shows that seepage water flows more easily into the ditch bottom than that of Figure 8 (a) when not considering backfill, because the backfills are more permeable than neighboring soils. On the other hand, in the case of the assumed condition as shown in Figure 8 (c), the seepage water flows into the ditch bottom, avoiding the less permeable backfill. This means that the creep length for Figure 6: Collapse of ditch border (2004/05/24, during puddling) (a) when not considering backfills (b) when considering backfills Figure 7: Conditions of the ditch near portion boiling occurred (2003/12/12) seepage flow increases, and as a result the stability against seepage failure increases under the assumed condition. It is also found, from Figures 8 (a), (b) and (c), that seepage flows are only limited to either side of the ditch and interactions with the other side s flow are small in this case. The second case of H = 0.456m on 2004/05/22 was also analyzed and the same criteria as above were applied. VI STABILITY AGAINST SEEPAGE FAILURE OF THE BOTTOM WITHIN THE DITCH The calculation procedure using the Prismatic failure concept was given in detail by Tanaka et al. (1996). The critical hydraulic head difference for damage is given as the hydraulic head difference at the onset of deformation in experiments, H y [Tanaka et al. 2005], which is equal to the theoretical critical hydraulic head 1546
difference, H PF, calculated using the Prismatic failure concept [Tanaka & Verruijt, 1999]. According to the Prismatic failure concept, stabilities against seepage failure of bottom soil within the ditch were analyzed. For the conditions of H = 0.212m on 2003/12/12 and H = 0.456m on 2004/05/22, the critical hydraulic head difference H PF and safety factor for seepage failure F s are calculated as shown in Table 2. In the case of H = 0.212 m on 2003/12/12, H PF is 0.416 m (F s = 1.961) when not considering backfill, H PF = 0.362 m (F s = 1.706) when considering backfill, and H PF = 0.682 m (F s = 3.216) when considering less-permeable backfill. It is therefore concluded that the ditch bottom is stable for seepage failure. On the other hand, in the case of H = 0.456m on 2004/05/22, H PF is 0.460m (F s = 1.008) when not considering backfill, H PF = 0.394m (F s = 0.864) when considering backfill, and H PF = 0.749m (F s = 1.641) when considering less permeable backfill, and it is concluded that the ditch bottom was near critical or was subject to boiling. The ditch bottom would be stable against boiling if the back fill was one order less permeable, as an assumed condition. Thus it is proved that a less-permeable backfill has an effect of increasing the stability. It is therefore foreseen that, in the cases of H = 0.600 m on 2004/05/16 (at a heavy rainfall) and H = 0.878 m (an estimated value) on 2004/05/24, safety factors were decreased even further and the ditch bottom would be subject to boiling and/or collapse. (a) when not considering backfills (b) when considering backfills (c) when considering less permeable backfills Figure 8: Equi-potentials (2003/12/12) Table 2: Critical hydraulic head difference H PF and safety factor for seepage failure F s (b) Considering backfill (c) Less permeable backfill Condition (a) No backfill (in fact) (on an assumed condition) Date (H) H PF (m) F s H PF (m) F s H PF (m) F s 2003/12/12 (0.212 m) 0.416 1.961 0.362 1.706 0.682 3.216 2004/05/22 (0.456 m) 0.460 1.008 0.394 0.864 0.749 1.641 VII EFFECT OF 2DC FLOW CONDITION IN STABILITY AGAINST SEEPAGE FAILURE OF SOIL On excavating a large area, as shown in Figure 9, the seepage failure of soil in front of a sheet pile wall is a problem in two dimensions, which is designated as two-dimensional flow (2D flow). In the excavation of soil between double sheet pile walls, as shown in Figure 10, seepage water concentrates into the soil twodimensionally from the outside, which is called two-dimensional concentrated flow (2DC flow). The twodimensional concentrated flow of water lowers the safety factor regarding the seepage failure of soil [Bauer, 1984; Tanaka et al., 2002]. The case treated here is a similar condition to the 2DC flow. To discuss the effects of the 2DC flow in stability against seepage failure of the ditch bottom, an assumed 2D flow (i.e. a single-sheet-pile-wall-type) and H = 0.212m on 2003/12/12 condition is analyzed when not considering a backfill as shown in Figure 11. From FEM seepage flow and stability analyses, H PF is 0.404 m (F s = 1.908) which is nearly equal to that of the double-sheet-pile-wall-type H = 0.416 m (F s = 1.961). 1547
Figure 9: Two dimensional flow (2D flow) Figure 10: Two dimensional concentrated flow (2DC flow) Taking the ratio, B/D, of a half width of double sheet piles, B, to the penetration depth of sheet pile, D, it is concluded by Tanaka et al. (2011a) that the condition of an excavation bottom should be treated as a 2DC flow when B/D 5, and can be otherwise treated as 2D when B/D > 5. In the case treated here, we have the condition of 2B = 1.140m, D = 0.089m, and B/D = 6.4 (>5), so it can be said that the effects of 2DC flow do not affect stability against seepage failure from a theoretical point of view. VIII COUNTERMEASURES AGAINST SEEPAGE FAILURE Figure 11: Single-sheet-pile-wall-type ditch assumed (2003/12/12) Let us consider the following three countermeasures against seepage failure: (1) increase in the penetration depth of the panel, (2) decrease in the coefficient of permeability of backfills, and (3) replacing the upper part of the bottom ditch soil with an inverted filter. Those methods are thought to be effective against seepage failure of soil in front of sheet piles. VIII.1 Increase in the penetration depth of the panel Consider the case of the planned section and H = 0.320m when not considering backfills as shown in Figure 13. Five cases of penetration depths D of 0.0, 0.09, 0.135, 0.18 and 0.28m were analyzed. An example of flow nets for the case of D = 0.090 m are illustrated in Figure 13. It is found from Figure 13 that seepage water goes around the bottom tip of the panel. Figure 14 Figure 13: Flow nets in the case of D = 0.09m 1.5 1.0 H PF (m) 0.5 Figure 12: Increase in the penetration depth of the panel D (planned section, H = 0.320m) 0.0 0.00 0.10 0.20 0.30 0.40 D (m) Figure14 : Relationship between D and H PF 1548
shows the relationship between the depth of the panel and the theoretical critical hydraulic head difference H PF. It is observed from Figure 14 that H PF efficiently increases with increase in D. H PF however does not increase more with increase in D beyond 0.18 m, because the less permeable silt layer appears for the depth D 0.18 m in this site. For the penetration depth of the panel D = 0.18 m, H PF is given as 1.343 m. For this case, the bottom of the ditch is therefore stable against seepage failure (F s = 1.530) even for the maximum hydraulic head difference H = 0.878 m estimated at this site. VIII.2 Decrease in the coefficient of permeability of backfills Let us consider the case of the planned section and H = 0.320 m when considering backfills as shown in Figure 15. Seven coefficients of permeability k BF of 2.265 10 5, 1.5 10 5, 1.0 10 5, 0.5 10 5, 0.25 10 5, 1.0 10 6 and 1.0 10 8 m/s were analyzed. From Figure 16 showing the relationship between k BF and H PF, it is found that the value of H PF increases and converges in the vicinity of 1.0 m with decreasing k BF. Lesspermeable backfill is also effective in increasing stability against boiling of the bottom soil. 1.2 1.0 0.8 H PF (m) 0.6 0.4 0.2 0.00 0.50 1.00 1.50 2.00 k BF ( 10-5 m/s) Figure 15: Considering backfills (planned section, H = 0.320m) Figure 16 : Relationship between k BF and H PF VIII.3 Replacing the upper part of the bottom ditch soil with an inverted filter Here consider an example of penetration depth of a panel D = 0.135m, and replacing the upper part (0.09m) of the bottom ditch soil with an inverted filter (i.e., soil that is more permeable than the soil at lower depths) as shown in Figure 17. Six cases of permeability of filter k F = 1.31 10 5, 2.62 10 5, 6.55 10 5, 1.31 10 4, 5.24 10 4 and 1.31 10 3 m/s are analyzed. Figure 18 shows the relationship between k F and H PF. It is observed from Figure 18 that H PF effectively increases with increase in k F. The H PF value is given as 0.729 m for k F = 1.31 10 5 m/s which corresponds to the case with no replacement, and is given as 1.390m for the coefficient (k F = 1.31 10 4 m/s) ten times as large as the above one. Thus replacing the upper part of the bottom ditch soil with an inverted filter is also effective in increasing stability against boiling. 1.75 1.50 H PF (m) 1.25 1.00 0.75 0.50 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 Figure 17: Bottom ditch soil replaced with an inverted filter (planned section, D = 0.135m) k F ( 10 4 m/s) Figure 18: Relationship between k F and H PF 1549
IX CONCLUSIONS Boiling occurred in the bottom soil within a double sheet-pile-wall-type ditch (or prefabricated ditch) during puddling, and the ditch border collapsed. The penetration depth was so limited that the bottom soil of the ditch became unstable when the groundwater level increased. Changes in the groundwater level were observed continuously from 2003/12/05 to 2004/04/03. The largest hydraulic head difference, H, during the continuous observation period was 0.212m on 2003/12/12. On intermittent observation, the H values ranged from 0.600m on 2004/05/16 after heavy rain to 0.456m on 2004/05/22 during puddling. The greatest H value at the time of transplanting was estimated to be 0.878m on 2004/05/24, when boiling failure occurred. FEM seepage flow and stability analyses were conducted, and the safety factors against seepage failure were calculated as F s =1.961 (under the conditions on 2003/12/12) and 1.008 (under the conditions on 2004/05/22) when not considering backfills, and as F s = 1.706 (2003/12/12) and 0.864 (2004/05/22) when considering backfills. The backfills were more permeable than neighboring soils in the site. It is concluded that the bottom soil within the ditch was stable under the conditions on 2003/12/12, but was near or beyond the critical condition for seepage failure under the conditions on 2004/05/22. In the case treated here, we have the condition of 2B = 1.140 m, D = 0.089 m, and B/D = 6.4 (>5). It is found that seepage flows are only limited to either side of the ditch and interactions with the other side s flow are small, and that the effects of 2DC flow do not affect the stability against seepage failure. In the case of long hydraulic structures such as a ditch treated here, geology can change along the length, so it should be confirmed that the stability is satisfied along the entire length. Three countermeasures against seepage failure were considered: (1) increase in the penetration depth of the panel, (2) decrease in the coefficient of permeability of backfills, and (3) replacing the upper part of the bottom ditch soil with an inverted filter. It was found that the first method was the most effective. When the penetration depth of the panel is D = 0.18 m, the bottom of the ditch becomes stable against seepage failure for the maximum hydraulic head difference H = 0.878 m estimated at this site. The combination of establishing less permeable backfills and replacing the upper part of the bottom ditch soil with an inverted filter would also be effective in increasing stability against boiling of the bottom soil. The reason why boiling did not occur even with a high ground water level 240 meters from the point was given by Tanaka et al. (2011b) as that the bottom soil of the ditch consisted of clay of low plasticity (CL). X REFERENCES Bauer, G.E. (1984). Dewatering, hydraulic failure and subsequent analysis of a sheeted excavation, Proceedings of International Conference on Case Histories in Geotechnical Engineering, 1415-1421. Tanaka, T., Hirose, T., Inoue, K. and Nagai, S. (2005). Performance-based design concept of soil for seepage failure within cofferdam. Transactions of the Japanese Society of Irrigation, Drainage and Reclamation Engineering, 73-6: 107-116. Tanaka, T., Hori, H. and Inoue, K. (2002). Boiling occurred within a braced cofferdam due to twodimensionally concentrated seepage flow. Proceedings of the 4th International Symposium on Geotechnical Aspects of Underground Construction in Soft Ground (IS-Toulouse 2002). 459-464. Tanaka, T., Shimizu, K., Shibata, S. and Inoue, K. (2011a). Non dimensional formulation of critical hydraulic head difference for seepage failure of soil in a cofferdam. Proceedings of the 46th annual meeting of Japanese Geotechnical Society (in Japanese): 1005-1006. Tanaka, T., Takashima, W. and Pham T. H. T. (2011b). Seepage Failure of Bottom Soil within Double Sheet-Pile-Wall-Type Ditch Stability against Heaving of Bottom Clay Soil in the Ditch where Boiling has not occurred. Report of Research Center for Urban Safety and Security, Kobe University, 15: 257-264. Tanaka, T., Toyokuni, E. and Ozaki, E. (1996). Prismatic failure A new method of calculating stability against boiling of sand within cofferdam. Proceedings of International Symposium on Geotechnical Aspects of Underground Construction in Soft Ground, 219-224. Tanaka, T. and Verruijt, A. (1999). Seepage failure of sand behind sheet piles The mechanism and practical approach to analyze, Soils and Foundations, 39-3: 27-35. 1550