Effectiveness of horizontal drains in improving slope stability: a case ...

2 FIELD STUDY2.1 Background of PutrajayaPutrajaya is a planned city, located 25 km south ofKuala Lumpur and 20 km north of the KualaLumpur International Airport (KLIA) that serves asthe federal administrative centre of Malaysia. Thecity sits on 4,931 hectares of land within theMultimedia Super Corridor. Its development startedin 1995 and is targeted to complete by 2015. Oncefully developed, it is estimated to have a populationof around 320,000 with 64,000 housing units.2.2 The Putrajaya landslideA landslide occurred at a slope adjacent to 3 blockscontaining 334 government quarters at Precinct 9,Putrajaya, at around 4:30am on the 22 nd March2007. The panaromic view of the landslide in Figure1, reproduced from the report that appeared inthe New Straits Times on 23 rd March 2001, providesgraphic illustration of the landslide, the location ofthe site and an indication of the distance from theslope to the adjacent apartment buildings.Figure 1. Panaromic view of the slope failure, (New StraitsTimes, 23 rd March 2007).There was no injury or fatality, but 27 cars were totallyor partially buried by the debris material. Allresidents of the 3 blocks, i.e. about 1,200 people,were immediately evacuated. It was therefore crucialto stabilize the affected slope prior to allowingthe residents to reoccupy the buildingFigure 2. Digital Terrain Model (DTM) of the landslide area.2.4 Geology of the siteFrom the field surface mapping, the study area isunderlain by graphitic quartz mica schist from KajangFormation (see Figure 3). The outcrops whichcould be noted along the cut slopes in the study areahave experienced intense weathering processes withweathering grade range from Grade III to VI. HoweverGrade II material could still be noted at certainareas. Foliation planes were clearly formed andcould be seen on the cut slopes. The orientation isquite consistent, i.e. striking approximately N-S anddipping towards W-EW with average dips of 35 o to40 o .Study Area2.3 Site description of landslide areaThe site is best described by the Digital TerrainModel in Figure 2. The landslide is located on thewestern face of a ridge that has its crest close to theupper edge of the slide scarp. A road leading to a 36million litre water tank lies on this crest. The watertank is partially buried on a part of the hill that hasbeen cut to prepare a flat site for it. The distance ofthe tank to the edge of the ridge where the slide occurredis about 25m.Kenny HillFormationKajangFormationFigure 3. Geology map of study area754

The outcrops (Grade II to IV) are dark grey toblack in color with clear foliation planes. For GradeV-VI outcrops, the color is reddish brown to brownishred due to existence of laterite which is rich inferrous oxides.The results from petrography of the core samplesat depths of 9 - 20 m, reveal that the rocks in thisarea consists of interbeded sandstone, shale and actinoliteschist.2.5 Soils investigation and instrumentationThe first soils investigation and instrumentationmonitoring package commenced in March and wascompleted in April 2007. 17 boreholes were sunkby rotary drilling with foam and water as flushingmedium (see BH1-BH17 in Fig. 4). When rock wasencountered, rotary drilling was carried out with diamond-studdedbit to obtain rock core.Seven standpipe piezometers were installed (seeSP1-SP7 in Fig. 4) to obtain the piezometric waterlevel readings, so that the performance of the dewateringscheme using horizontal drains could bemonitored. The standpipe piezometer, or sometimescalled a Casagrande piezometer, consists of a filtertip joined to a PVC riser pipe. After the filter tipand riser pipe were installed in the borehole, a sandfilter zone was tremied into place around the filtertip. The top of the filter zone is sealed with bentoniteto isolate the pore water pressure at the tip.The riser pipe was terminated above ground levelwith a vented cap. Readings were manually takenonce to three times a day using a water level indicator.Five vertical inclinometers were also installed(see IC1-IC5 in Fig. 4) to detect zones of movementand determine whether the slope is responding to theimmediate remedial measures. Inclinometer casingswere installed permanently in boreholes. A portabletraversing probe and a readout unit were used tomeasure and record the movement at 0.5 m depth interval.2.6 Horizontal drains installationBased on field data, preliminary slope stabilityanalysis were carried out which indicated that theprobable cause of failure was a rise in ground watertable within the slope. As a result, deep horizontaldrains up to 24 m, instead of the normal 6-12 mlength drains, were immediately installed at or nearthe toes of the intact slopes.A total of 40 horizontal drains ranging from 8 to24 m in lengths were installed from 28 th March to 1 stApril 2007 (Fig. 5). They were constructed by drilling100 mm diameter holes at an upward incline ofabout 5 degrees from the base of the slope and theninserting a 75 mm nominal diameter class D perforatedPVC pipe wrapped in geotextile into the drilledholes. The total drilling length was 842 m while thelength of pipe installed was 784 m.Thirty-seven drains were installed at the toe belowthe lowest berm, while another three were installedjust above the first berm. Figure 5 shows thelayout of the horizontal drains installed. It was, surprisingly,observed that the yield of the drains,measured in ml/s, greatly varies from one to another.Figure 4. Locations of boreholes and instrumentationFigure 5. Horizontal drains locations, and resistivity and seismicsurvey lines755

Soon after installation in 2007, all 40 horizontaldrains, except for 2, were able to discharge water.The total discharge rate measured from all drains, atthat time, was 652.19 ml/s. However, 71% of theyield comes from 5 drains (listed in the Table 1)close to the northern side of the failure area. Flowrates measured from the other 35 drains range from0-18.02 ml/s, with 31 of the drains yielding less than10 ml/s.These observations concurs with Fernandez-Rubio and Singh (1983) which discovered yield ofhorizontal drains from dry to very high flow rates,with the maximum discharge located in the areas offault zones or slope failures.The flow rates measured after 3 years were muchless than immediately after installation. Only theflow rates from 4 horizontal drains were able to bemeasured. The total yield was only 114.56 ml/s,17.5% of the yield immediately after installation.Although the flow rates from horizontal drainshas much reduced after 3 years, the peizometric waterlevel measured on 8 th March 2011 at BH5/SP2, 4years after the drains were installed, was 5.25 m,about the same level it used to be, immediately afterthe installation of the drains, in April 2007.Five inclinometers were installed to monitor lateralmovement of the slope. Except for BH2/IC2which is located at mid height next to the landslide,there is very little movement of the rest of the slopesince the inclinometers were installed soon after collapse.Figure 6 plots the maximum movement detectedat 2.5m below the ground surface at this inclinometer.It can be seen that there’s about 8mmmovement during the horizontal drains installationworks, after which the inclinometer cease to showany.The above observations suggested that horizontaldrains is an effective method to lower down groundwater table and improving slope stability in the shortand medium term. Further studies are, however, requiredto determine its effectiveness in the longterm.3.2 Results from geophysical surveysThe topography model for the seismic line S1 issimplified in Figure 7. Three seismic boundarieswere detected. The first topmost layer is characterizedby the lowest p-wave velocity with values rangingfrom 260 m/s to a maximum value of about 500m/s. These values represent the range of seismic velocitiesof the inhomogeneous near surface soil materials.The second layer is particularly exhibiting lowervelocities of 650 to 1300 m/s. On the basis of theirreflection configurations, this layer could beassociated with highly weathered rock materials orstiff/hard sediment.The third layer is interpreted as the less weatheredrock mass. The computed seismic velocities rangefrom 1926 to 4350 m/s. Theoretically, the low valueof seismic velocity is associated with highlyweathered bedrock, whereas the higher valuerepresents relatively less weathered bedrock. Hence,the third seismic unit could be interpreted as highlyand less weathered fresh Schist.Figure 7. Seismic survey result for line S1Figure 8 shows the 200 m resistivity measurementswith 5 m electrode spacing of survey line R1 (Fig.5). The top layer of the pseudo section is interpretedas a dry layer of silt and clay with low resistivity inthe range of 10 to 1,000 Ωm. The relatively high resistivityin the range of 5,000-30,000 Ωm observedin second layer comprises of slightly weathered tofresh schist. Rock Quality Designation (RQD) of70%-90% at a depth of 11.5m from BH1 substantiatesthis.Figure 8. Resistivity survey result for line R1757