Construction of a 12m High Embankment in Hydraulic Sand Fill

Transcription

1 Construction of a m High Embankment in Hydraulic Sand Fill W. H. Ting Consulting Engineer, Malaysia tingwh@pop.jaring.my D. E. L. Ong Ph.D Research Scholar, National University of Singapore, Singapore Formerly Jurutera Jasa (Sarawak) Sdn. Bhd., Malaysia engp9@nus.edu.sg L. Y. Tai Director, Jurutera Jasa (Sarawak) Sdn. Bhd., Malaysia tai@juruterajasa.com A. T. C. Wong Regional Manager (Western Region), Hock Peng Organisation, Malaysia alexwong@hockpeng.com Abstract: A m high embankment that formed part of an approach to a large span river bridge is to be constructed with hydraulically placed sand-fill. In the course of construction, toe failure of the embankment took place. The construction of the embankment was suspended but it was decided to continue with the same construction with an improved design of the embankment. This paper describes in detail the results of the seepage and slope stability analyses as well as site monitoring of the phreatic surface during construction of the re-designed embankment. The re-designed embankment was successfully constructed to its full height of m without further incidents. 1 INTRODUCTION A second crossing over Sg. Sarawak linking Petra Jaya and the city of Kuching, Malaysia was to be constructed to ease the traffic flow on the Satok Bridge crossing located at the upper reaches of the river. A m high hydraulic sand-fill embankment that formed part of the approaching section to the large span bridge structure was to be built. During construction, embankment toe failure occurred and construction had to be suspended. Detailed transient seepage and slope stability analyses have been carried out to investigate the causes of the toe failure. Based on the detailed analyses, an improved embankment design was implemented. An instrumentation program that included water standpipes was also implemented to assist in monitoring the stability of the embankment to prevent further incidents. The proposed hydraulic sand-fill embankment was located in a valley. The hydraulic sand-fill was pumped from a nearby river and had noticeable silt content. In locations where some fines were present, the fines were successfully transported away as part of the surface water drained away from the hydraulic fill. A layout of the approach embankment section is shown in Figure 1. The foundation of the embankment was in alluvium. As foundation treatment, the weaker natural overlying strata was excavated and replaced with sand. During construction, there were no signs of basal movement and hence bearing capacity of the foundation was not in question and was expected to be able to support the embankment. PROPOSED EMBANKMENT AND SUBSURFACE CONDITION Figure 1. Proposed plan view of hydraulic sand-fill embankment. 1

2 3 PROBLEM ENCOUNTERED When the embankment reached m in height, toe failure occurred. Subsequent hydraulic sand-filling work was temporarily suspended to prevent more toe failures from occurring at different locations of the embankment. From site inspection, the toe failure was believed to be caused by wash water transporting the hydraulic sand-fill at the top of the embankment at that stage of the construction, seeping downwards through the embankment sand-fill and building up in the embankment. The possibility of rapid and considerable build up of pore water pressure within the embankment during hydraulic sand-filling therefore needs to be investigated, and a remedial proposal is to be recommended. REDESIGN CONCEPT As the sand-fill was hydraulically placed, an unconfined seepage flow within the embankment was anticipated. Such unconfined seepage flow is typically a boundary value problem, whose seepage boundary is unknown and has to be determined as part of the solution (Cividini & Gioda (19), Desai & Li (193) and Bathe et al. (19)). The phreatic surface that develops within the embankment during hydraulic sand-filling can be modelled using a commercially available software called SEEP/W. Once the phreatic surface has been determined, it may then be exported to a related software called SLOPE/W so that stability of the embankment shoulder can be assessed. The aim of modeling the embankment using SEEP/W and SLOPE/W is to establish an improved embankment design that would take care of drainage during construction as well as in the finished embankment. MODELING PROCEDURE AND RESULTS Since the cross-section of the embankment is symmetrical, only half of it is modeled in SEEP/W. The left and right foundation boundaries are deliberately modeled as steep slopes to prevent compatibility problem when the SEEP/W finite element mesh is exported to SLOPE/W, since the latter does not permit input of vertical segments. The proposed channel transporting the hydraulic sand-fill is modeled at the middle of the crest of the embankment. The depth of the channel is approximated to be about 3m and the hydraulic sand-fill is expected not to exceed a height of m at all times. From site inspection, it was considered that the toe failure was caused by rapid build up of pore water pressure within the embankment during hydraulic sand-filling. Therefore, proper drainage is an important design consideration. In view of this, it is proposed that a highly permeable rock toe with an extended berm of m be constructed at the toe of the existing embankment. This design consideration is modelled in SEEP/W. The input soil properties for the SEEP/W and SLOPE/W analyses are found in Tables 1 and, respectively. The partially saturated compacted sand-fill can be modeled using volumetric water content functions so that transient analyses can be performed. Due to unavailable field data, the volumetric water content function from Ho (1979), which can be found in SEEP/W database, is used for this hydraulically placed sand-fill. The particle size distribution and saturated conductivity from Ho (1979) is reasonably similar to that of the sand-fill. It is believed that an approximate function (due to unavailable of field data), should lead to more realistic results than using a single-value function when dealing with unsaturated flow. From site observation, it was noted that the deposited hydraulic sand-fill would take about 1 hour to reach a height of m in the approximately 6m long channel. Likewise, the drawdown rate of the channel was also noted to be about m/hour from the onset of suspension of sand-filling. Table 1. Input parameters for seepage analyses. Material type Permeability (m/s) Volumetric water content (Porosity x Saturation) Hydraulic fill 5. x Stone toe 1 x Table. Input parameters for slope stability analyses. Material type Effective angle of friction (deg) Cohesion (kpa) Hydraulic fill 37.5 Foundation 1 6 Foundation 5 These rates are thus used as boundary conditions for the filling and drawdown of the channel in the SEEP/W transient analyses. An appropriate time interval for the transient analyses is used to capture the fluctuations of the phreatic surfaces during and after the sand-filling process. The pre-determined phreatic surfaces are then exported to SLOPE/W so that slope stability analyses can be performed. 5 DISCUSSION AND FINDINGS Wiesner () reported that the weight of an embankment built on soft subsoil would cause a concave settlement profile. Due to the differential settlement underneath the embankment, vertical stresses are expected to reduce in a zone along the centerline of the embankment and be transferred towards the toe. As such, the horizontal shear stresses will also increase towards the toe. This could cause the toe of the embankment to weaken. Results of the seepage analyses reveal that the construction of the embankment to a height of m would also result in relatively high exit gradient and horizontal seepage velocity (see Figure ) at the embankment toe. It is also thought that some of the water from the hydraulic sand-filling process flowed downwards and in areas of high ground water table level, the flow at the base of the embankment was in the direction towards the toe. It is known that under horizontal seepage conditions, the limiting angle of slope is φ /, where φ is the friction angle of the sand-fill. The limiting angle for the site material is therefore about 37.5/ =1. o and since the embankment slope angle is around 6.6 o, it should not be surprising that the embankment toe is susceptible to limiting conditions. Therefore, in view of the possibility of stress concentration and horizontal seepage velocity at the embankment toe, failure is to be guarded against in the toe area. In order to cater for the toe horizontal seepage and to increase the strength of the embankment toe against sliding due to the high horizontal shear stresses, it is recommended to construct a permeable and rigid rock toe wrapped with geotextile. The geotextile would prevent sand from being washed into the voids within the rock toe.

3 Figure. High exit gradient and horizontal seepage velocity at toe could be the culprit behind the toe failure during construction. Figure 3(a) shows a simulated embankment, whose height (H) and width from the side of the flow channel to the edge of the slope (W) are m and m, respectively. It is observed that after just 3 minutes of continuous sand pumping, the phreatic surface rose rapidly so that it intercepts the slope surface at embankment height of about. m from the existing ground level. This phreatic surface is then exported to SLOPE/W so that slope stability analyses can be performed. The result reveals that the embankment shoulder yields a factor of safety of just.96 against failure, as shown in Figure (a). The factor of safety after hours of continuous sand-filling is noted to have reduced to., as shown in Table 3. After 3 minutes of pumping After hours of pumping m 5 m m 7 m Figure 3(a). Development of phreatic surfaces due to hydraulic sand-filling after 3 minutes and after hours, respectively. After 3 minutes of pumping After hours of pumping m 5 m m 7 m Figure 3(b). Development of phreatic surfaces due to hydraulic sand-filling after 3 minutes and after hours, respectively. The flow channel for hydraulic transport of material at the crest of the embankment is shown in Figure 5. If W is increased to m for the same embankment height, 3 minutes of continuous sand-filling will produce a phreatic surface as shown in Figure 3(b). Using this phreatic surface as input to SLOPE/W, a higher FOS of 1.1 is obtained as shown in Figure (b), as opposed to.96 if W is m. Therefore, it is evident that the FOS can be increased by lengthening the flow path from the flow channel to the rock toe of the embankment. However, after hours of continuous sand-filling, the FOS would reduce to.9 (see Table 3), which shows that the critical phreatic surface has been reached. As such, the sand-filling process must be suspended so that the critical phreatic surface is allowed to drop to a safer level before the sand-filling process is allowed to resume. After hours of interruption 5 m 7 m After hours of pumping m m Steady-state Figure 3(c). Drop of phreatic surface due to suspension of sandfilling for hours. Figure (a). FOS after 3 minutes of hydraulic sand-filling (H=m, W=m). - - W.956 H H Figure (b). FOS after 3 minutes of hydraulic sand-filling (H=m, W=m). W 3

4 Table 3. FOS of embankment shoulder against failure during hydraulic sand-filling Embankment Duration of continuous FOS Configuration (m) sand-filling 3 mins.96 H=, W= hours. H=, W= 3 mins 1.1 hours.9 After suspension for hrs 1.7 After suspension indefinitely 1. From the results of the transient analyses, it is found that if the hydraulic sand-filling process is suspended for at least hours, the phreatic surface would drop to a safer level, as illustrated in Figure 3(c). By suspending the sand-filling process for hours, the corresponding FOS is noted to have increased to 1.7 as shown in Table 3. (d) Implementation of water standpipes in the embankment to monitor the fluctuation of phreatic levels due to (c) above. If the head difference between phreatic surface and ground level is.5 m or lesser, hydraulic sand-filling must be suspended immediately. Figure 6. Improved design of embankment toe. Figure 6 shows the improved design for the embankment toe, which consists of a 1m high triangular rock toe with a.7m base. The rock toe has a.3m thick and 1.9m wide berm extending into the hydraulic sand-fill. The rock toe is to be wrapped with geotextile to prevent sand particles from in-filling the voids within the rock toe. In addition, the shoulders of the embankment would be protected by a nominal earth fill layer and turfed to prevent erosion of the hydraulic placed fill during downpours. 7 SITE OBSERVATIONS AND MEASUREMENTS The recommended re-design was implemented (see Figure 7) and construction of the embankment resumed. Figure 5. Flow channel for material transport at crest of embankment. A -hour interval of hydraulic filling-suspension is also practical from the construction view point. As a comparison, if the hydraulic sand-filling process were to be suspended indefinitely, the phreatic surface would approach steady-state condition as shown in Figure 3(c). The steady-state is to simulate persistent rainfall and total suspension conditions as shall be elaborated in the next section. When this occurs, the FOS would increase further to ADOPTED CONSTRUCTION METHOD AND DESIGN From the analyses above, it was recommended that construction of the new and improved embankment adhere strictly to the following construction methods: (a) Width from the side of the flow channel to the edge of the embankment shoulder should be at least m. (b) Rock toe with its extended berm to be constructed at the base of existing embankment. (c) Hydraulic sand-filling process to be suspended every hours to allow drop of phreatic surface so that the FOS against embankment slope failure can be increased. Figure 7. Construction of rock toe wrapped with geotextile. Water standpipes were installed in the sand-fill embankment to monitor the fluctuation of the phreatic surface during sandfilling. The instrumentation program was carried out to ensure that the design criteria based on the computer simulation were within working limits to prevent embankment slope failure. It so happened that the construction of the highest section of the embankment (m) coincided with the year-end monsoon where heavy downpours were normal. Therefore, this period of construction was considered the most critical and the field measurements from 1-3 Nov at

5 Chs. 3 and 35 (see Figure 1) were used for illustration and comparison. Figure (a) shows the data from the SEEP/W analysis presented in Figure 3(c). It is observed that the interruption of pumping of hours would cause uneven drop in phreatic surface along the section of the embankment. The phreatic surface would not reduce much for locations nearer to the toe of the embankment, despite showing relatively larger reduction nearer to the centre of the embankment. This highlights the susceptibility of embankment toe and shoulder failure. If pumping is suspended indefinitely with an assumption that persistent rainfall would sustain a steady head, a much lower steady-state phreatic surface would be achieved. Water head (m) Drop in phreatic level (m) Distance from embankment toe (m) Figure. (a) Extracted water head data from SEEP/W analysis and (b) comparison between computed and measured drop in phreatic level. Figure (b) shows the computed and measured drop in phreatic levels. It is observed that the computed data for distances 7 m and m away from the embankment toe generally gives values that are about the average of the measured values. However, at 11 m from the embankment toe, the computed drop in phreatic level is slightly higher than the average measured values. From Figure (b), it is also noted that the drop in phreatic levels increases with distance away from the embankment toe, similar to prediction made by SEEP/W as discussed earlier. The discrepancies between the computed and measured values could (a) After hours of pumping After hours of suspension of pumping After indefinite suspension of pumping (b) Drop after hours of suspension of pumping Drop after suspension of pumping indefinitely Measured drop be due to persistent rainfall that might have prevented the phreatic surface from dropping further. For this reason, hydraulic sand-filling had to be suspended immediately on Nov and Nov as the head difference between the phreatic surface and ground level was less than.5m. From both the computed and measured data, it is evident that if the toe of the embankment is not strengthened and proper drainage paths are not provided, the embankment toe can easily fail due to pore water pressure buildup. Therefore, when the pore water pressure buildup is reduced by introducing an improved embankment design (Figure 6) and proper construction method, the embankment was successfully constructed without any further incidents. CONCLUSION The toe failure of the hydraulic sand-fill embankment was believed to be caused by wash water transporting the hydraulic sand-fill at the top of the embankment at that stage of the construction, seeping downwards through the embankment sand-fill. The construction was suspended for an investigation to be carried out on the possibility of rapid build up of pore water pressure within the embankment. Detailed seepage and slope stability analyses have been performed to determine proper construction method and to establish an improved embankment design so that construction of the embankment can be continued to its maximum height of m. A field instrumentation program consisting of water standpipes was implemented during the construction to consistently monitor the fluctuations of the phreatic surface so that the construction method derived based on the computer simulation were within working limits to prevent further failure. It has also been illustrated that the computed and measured water levels generally show fair agreement. It was believed that due to persistent rainfall, the drop in phreatic surface was not as great as predicted for suspension of pumping operation; especially at locations near to the embankment toe despite suspension of sand-filling. Nevertheless, the development of phreatic surface had been successfully controlled on site by monitoring the water standpipe readings and on a few occasions, pumping had to be terminated immediately due to high phreatic surface approaching the embankment shoulder. The m high hydraulic sand-fill embankment had been successfully constructed without any further incidents after the proposed revised construction method had been strictly adhered to. REFERENCES Cividini, A. and, G. 19. Approximate finite element analysis of seepage with a free surface. Int. Journal for Numerical and Analytical Methods in Geomechanics, Vol. (6), pp Desai, C. S. and Li, G. C A residual flow procedure and application for free surface flow in porous media. Advances in Water resources Research, Vol. 6(1), pp Bathe, K. J., Sonnad, V. and Domigan, P. 19. Some experience using finite element methods for fluid flow problem. Proc. of the Fourth Int. Conference on Finite Element Methods in Water Research, Hannover, pp. 3-. Ho, P. G The prediction of hydraulic conductivity from soil moisture suction relationship. MSc. Thesis, Department of Civil Engineering, University of Saskatchewan, Saskatoon, Canada. 5

6 Wiesner, T. J.. Failure stresses beneath granular embankments. Developments in Theoretical Geomechanics, Sydney, Australia, pp. 33-, Smith and Carter (Eds), Balkema, Rotterdam. 6