Paper No.

536

FAILURE OF MECHANICALLY STABILIZED EARTH WALL AT KM 18 OF NH 6 A CASE HISTORY

D. ROY1, R. SINGH2, C.C. BHATTACHARYA3, D.P. GHOSH4 & S.P. DASGUPTA5

SYNOPSIS
A Mechanically Stabilized Earth (MSE) highway bridge approach, about 11 m in height, near Kolkata, failed recently within a month of its opening for traffic. Fortunately, direct collateral damage from the incident was small. The structure was founded on soft and compressible, fine grained soils of the intertidal flats and backswamps of the Ganges delta. Available geotechnical data indicated that design inadequacy was the main cause of the failure. Using pre- and postconsolidation shear strengths the embankment was redesigned. Possible remedial measures for the reconstruction of the embankment were explored. The selected reconstruction scheme involved installation of prefabricated vertical drains (PVDs) underneath a stabilizing berms running along the unstable segment of the MSE wall. The facilities are now operational and appear to be performing satisfactorily. Details of the failure, post failure investigations and monitoring, and redesign are presented in this paper.

1. INTRODUCTION

Highway embankment failed recently in the southern part of West Bengal at km 18 of NH6 Photo 1. The site is located within the intertidal flats and backswamps of the Hooghly, a major distributary of the Ganges. The area is underlain mainly by fine-grained silts and clays of Holocene and Pleistocene age (Vaidyanadhan and Ghosh1).

Essential details of the incident are presented first followed by the outlines of the subsurface investigation, monitoring programs undertaken in the post-failure stage, redesign of the earth structure, and subsequent reconstruction work. The authors were associated with the project after the failure.
2. INCIDENT

Photo 1. Southward view of the failure at km 18 of NH6

The site remains waterlogged throughout the year. The embankment was retained by Mechanically Stabilized Earth (MSE) wall constructed with compacted river sand reinforced with galvanized steel reinforcements.

The failure involved an MSE wall that runs along the western edge of the northbound lanes of a highway approaching a railway overpass. The failed MSE wall was constructed for four laning of a pre existing two lane, undivided highway that runs on the 30 year old embankment with 3 (horizontal) to 1 (vertical) side slopes along the eastern edge of the approach carrying southbound traffic. The incident occurred immediately after midnight on February 9, 2006 about a month after the highway was opened for vehicular traffic. During the incident, a section of the newly constructed 2-lane approach vertically settled by about 3 m and laterally translated outward by about 1 m. Although the movement was relatively rapid initially, post failure records indicate that the structure continued to move for nine days following the initial trigger (Fig. 1). The site plan is also shown on Fig. 2. The MSE wall appeared to have failed due to external instability without significant internal distress.

1. Deptt. of Civil Engg. IIT Kharagpur, Email:debasis@civil.iitkgp.ernet.in 2. Research Scholar, Deptt.of Civil Engg. 3. Former D.G. (R.D.) MOST., Consultant, Stup Consultants P. Ltd., Kolkata. 4. Professor, Deptt. of Civil Engg. IIT Kharagpur. 5. Professor, Deptt. of Civil Engg. Written Comments on this Paper are invited and will he received upto 31 st December 2007.

194

ROY, SINGH, BHATTACHARYA, GHOSH & DASGUPTA ON on the original geotechnical investigation and design of the highway in this area can be found in (Gangopadhyay and Das2). In the recent constructions for the four laning project, the north approach to the railway overpass at km18 site, that remained stable since its construction, was of similar details as those of the failed MSE wall. The subsurface conditions at the north approach site and the location of km 18 failure were also comparable. However, unlike the failed MSE wall, the north approach was constructed on ground improved by PVD installation followed by preloading. Incidentally, in early 2005 a 10 m high highway interchange MSE structure at km 26 of NH6 also failed because of external instability when it was in the final stages of its construction. The subsurface condition at km 26 is similar to that at km 18 site. However, the MSE embankment was constructed over foundation soils treated with PVDs. The highway interchange was reconstructed by providing an external earth buttress (berm) along the outer face of the MSE wall. The reconstructed interchange is carrying vehicular traffic for the last 22 months and appears to be performing satisfactorily.
4. SUBSURFACE INVESTIGATION

Fig. 1. Cross section

Fig. 2. Site layout and borehole locations

The failure received wide public attention because it occurred soon after the highway became operational. If such incidents are not reported, opportunities to learn from the occurrences are lost.
3. PAST HISTORY

The construction of the old embankment also triggered several slope failures. One such incident occurred in 1966, when the embankment reached its full height of 10.7 m. The instability was assessed to be due to a deep seated circular slip that day lighted just beyond the toe. Following this incident, the maximum embankment height within the stable stretch was 6.7 m. As remedial measures, the failed embankment was removed, the highway elevation was lowered and the embankment was reconstructed along with a 2.1 m high stabilizing berm along the edges of the embankment. Another slope failure occurred later immediately to the south of the railway tracks. Since poor subsurface conditions did not allow construction of an embankment of required height at the location of this failure, the earth embankment at the location of this failure was replaced by a structural viaduct for carrying the overpass. Details

Pre-failure geotechnical data basically originated from two boreholes, BH8 and BH5 (Figure 2). Other boreholes shown in Figure 2 were drilled after the failure at the instance of the authors since pre-failure subsurface data from only two boreholes, about 85 m apart, was considered insufficient to draw any inference regarding the nature of the underlying soil within the failure zone. The field work in these investigations included conducting SPTs and VSTs, and extraction of thin tube samples. The laboratory tests included UU triaxial and one dimensional incrementally loaded consolidation tests of selected thin tube samples, and tests for grain size distribution, natural moisture content, liquid limit and plastic limit. Data from these investigations indicate that the site is underlain by a sequence of Holocene silty clay, over silty clay of Pleistocene age. The upper 5 to 8 m of the Holocene unit was firm and over-consolidated with pre consolidation pressure up to 200 kPa, underlain by a 3 to 5 m thick soft, compressible, normally to lightly overconsolidated layer containing organics and a peat layer.

HIGHLIGHTS OF THE 178TH COUNCIL MEETING FAILURE OF MECHANICALLY STABILIZED EARTH WALL AT KM 18 OF NH 6 - A CASE HISTORY The deepest part of the Holocene unit was firm and normally consolidated. The Pleistocene unit contained sand or sandy silt partings and was stiff and overconsolidated with pre consolidation pressure up to 300 kPa. The soil samples classified as CL according to BIS 3 . One dimensional incrementally loaded consolidation test data indicate that at the time of failure the native soils were completely consolidated at the time of failure under the stresses imposed by the old embankment. In addition, the upper 2.5 m of native soils also likely to have undergone 50 per cent consolidation under the newly constructed MSE embankment and the stabilizing berm, while deeper native soil layers did not consolidate appreciably under the stresses imposed by these recent constructions. The estimates of undrained shear strength for the soils underneath km 18 site obtained from UU tests and VSTs are plotted in Fig. 3. The sensitivity of the soil layers were between 4 and 7.

195

construction. Dewatering began in late January 2004, after which the site was stripped to a depth of about 500 mm. A 500 mm thick compacted sand pad was placed on the stripped surface. Where the base of the MSE wall was to be at an elevation lower than the original ground level, the original ground surface was excavated for accommodating a 500 mm thick compacted sand pad underneath the base of the MSE wall. Sand filling commenced in early February 2004 and MSE wall construction began by the end of February 2004. The MSE wall and the reinforced sand embankment behind it were about 8 m high by mid June 2004 and 9.25 m by early February 2005. There was virtually no earthwork between mid June 2004 and mid January 2005. Earthwork began in end of November 2005 and the embankment construction was complete in early December 2005. After the completion of the paving work, the stretch was opened for vehicular traffic by mid January 2006.
6. EXTERNAL STABILITY REASSESSMENT

The failure pattern of the site is indicative of deep seated failure caused by external instability of foundation soils. Internally the MSE walls appeared to have been stable. The embankment was designed according to the Indian Roads Congress (IRC) document IRC: 754, which calls for a limit equilibrium factor of safety against external failure of 1.25 and an allowable settlement up to 600 mm with a provision that a smaller factor of safety may be acceptable in situations where the shear strength is expected to increase with time because of consolidation. Brief account on limit equilibrium stability reassessment of the original configuration of the MSE embankment is as follows. Data from the site indicate that (a) the undrained shear strengths at this location decreases westward and (b) the undrained shear strengths can be expressed as functions of effective vertical stress as shown with the solid and dashed lines in Fig. 3.
Fig. 3. Undrained shear strength, su

5. PRE-FAILURE CONSTRUCTION DETAILS

A brief chronology of the initial MSE wall construction is as follows. Since the site is permanently waterlogged, a 1-m high earth embankment was constructed in mid January 2004 about 1-m to the west of the MSE wall alignment for dewatering the area for

External undrained stability of the configuration of the MSE wall was assessed using the Software package XSTABL Version 5.1 (Interactive Software Designs, Inc.5) with simplified Bishop method (Bishop6). The input parameters used in the analysis were based on the following assessments regarding the degree of consolidation at the time of failure: The native soil layers underneath the MSE wall

196

ROY, SINGH, BHATTACHARYA, GHOSH & DASGUPTA ON

7. LESSONS LEARNT

were fully consolidated under the vertical stress imposed by the old embankment The degree of consolidation due to the MSE wall and the newly constructed embankment was 50 per cent over the top 2.5 m of native soils The deeper layers did not consolidate under the stress imposed by the MSE wall.

7.1. Undrained Shear Strength The main factor contributing to the instability of embankment appears to be in the estimation of post consolidation undrained shear strength of native foundation soils. The estimates of undrained shear strength ratio, su v , where is the undrained shear strength and v is the effective vertical stress, from km 18 site are plotted against the values of the overconsolidation ratio, OCR in Fig.5. The values of were estimated from incrementally loaded one dimensional consolidation tests conducted in the laboratory on thin tube soil samples. Also shown on Fig.5. is the range of correlations presented by Ladd et al.7 from five soft soil sites illustrating the SHANSEP framework. In spite of the scatter in the data, an approximate agreement between the data from the site and the correlations developed by Ladd et al.7 is apparent from the plot. The data also indicate that the undrained shear strength profile of the floodplain deposits found at the site is approximated by
0.76

The strength parameters for the old embankment was back figured from stability analysis to match the observed deep seated rotational instability during the construction of the embankment in the sixties described earlier. In this computation, the strength parameters for native foundation soils were assumed in accordance with the dashed line of Fig. 3. The results of external stability assessment for the MSE wall at failure indicate that the MSE wall was indeed marginally stable at the time of failure (Fig.4). The input parameters used in the computation are also included in Fig. 4.

(r2 =0.80).

s 0.25 (OCR ) that the SHANSEP approach is well suu v =Considering

known among geotechnical designers, overestimation of design strengtha major cause of failure at km 18 was probably because the inability on the part of the designers to calibrate the SHANSEP framework due to inadequate characterization of native foundation soils at the predesign stage.

Fig. 4. External stability reassessment

In comparison, the minimum factor of safety against circular slip in the original design under static loads was 1.42. However, in the original design the embankment height including pavement structure was assumed to be 8.4 m (against the actual of 10.5 m) and a total unit weight of 15 kN/m3 was assumed (in stead of 18 to 19 kN/m3 representative of the compacted river sand actually used to construct the embankment) for the embankment body because of the initial plan of constructing the embankment partly with fly ash. Also the undrained shear strengths used in the original design, shown on Fig. 3 for comparison, were greater than those inferred from laboratory and in situ test data over a large portion of the depth range of interest. Thus the bases of the original design were unconservative.

Fig. 5. Undrained shear strength OCR relationship

7.2. Trigger for External Instability As indicated earlier, the failure appears to be due to the use of inappropriately high undrained shear strength,

HIGHLIGHTS OF THE 178TH COUNCIL MEETING FAILURE OF MECHANICALLY STABILIZED EARTH WALL AT KM 18 OF NH 6 - A CASE HISTORY incorrect cross sectional geometry and unit weight of fill material used in construction in the initial stability assessments. Moreover, the embankment was only marginally stable at the time of the failure. No direct evidence of accelerated settlement rates immediately before failure is available for the site because of the absence of monitoring data covering the construction and early operational phase of the MSE wall. However, records from the site immediately after failure indicates that the failure was progressive in nature (Fig.1). As deformations increased, the shear strains within the native foundation soils likely to have exceeded those at which the peak undrained shear strengths are mobilized leading first to strain softening and then eventually to failure. Thus, the embankment failure may have been due to ongoing deformations of sensitive foundation soils and an inadequate factor of safety against external instability. In short, post failure investigations and review of pre failure records from the site indicate that incident could be avoided by undertaking adequate geotechnical investigation before design and appropriate coordination between the design office and construction activities.
8. RECONSTRUCTION ALTERNATIVES

197

1500 kNm. This alternative was not selected because of (a) difficulties in contract administration due to non availability of item rates on piling work and (b) relatively high cost of this alternative. 8.2. Lightweight Fill This option involves reconstruction of the embankment with light weight fills, e.g., expanded polystyrene (EPS) and foam concrete with a total unit weight of 3 kN/m3 and = 40 kPa. The static, short term, limit equilibrium factor of safety (simplified Bishop) in this option was 1.45 and the yield acceleration was 0.09 g. Thus, this scheme met the static and seismic external stability requirements. Requirement of 6700 m3 of light weight fill was estimated for embankment reconstruction. The time needed for reconstruction in this option was the least among all other options considered here. This scheme was not selected because of the (a) difficulties in procuring EPS blocks locally and (b) administrative constraints due to non availability of item rates of EPS blocks and their installation. Nevertheless, in many countries high embankments are often constructed economically with lightweight fill routinely in situations where the subsurface conditions are similar to those at the km 18 site. 8.3. Remedial Measures Based on PVD The other possible remedial measures at the site was explored with the installation of PVDs to a 13 m depth, construction of a two-stepped stabilizing berm along the outer face of the failed MSE wall, and reconstruction of the MSE wall and the highway embankment. This scheme was preferred because of (a) relative economy, (b) ready availability of PVDs and (c) availability of rates in the contract on the work items in this option. For the configuration shown in Fig. 6, the minimum factor of safety against overall rotational failure

8.1. Remedial Measures Based on Bored Piles Stability analyses indicated that the static factor of safety against external instability of the RE wall could be increased to 1.31 (simplified Bishop) rapidly by installing bored piles along the outer edge of the stabilizing berm that runs along the west of the failed MSE wall. It may be noted that IRC:75 calls for a static limit equilibrium factor of safety against deep seated external instability of 1.25. The yield acceleration (the pseudo static horizontal seismic coefficient at impending external instability) was 0.07g, which exceeded the design horizontal seismic coefficient 0.06 g used in this project. Thus, the scheme met the external stability requirements under static as well as seismic loading conditions. The scheme involved installation of three rows of 1-m diameter bored piles foundation up to 25-m depth. The horizontal spacing between piles within a row was 2.5 m and the row spacing was 3.1 m. The piles within the middle row were assumed to be staggered horizontally by 1.25 m with respect to the piles in the inner and outer rows. The piles were designed for an ultimate shear capacity of 375 kN and an ultimate moment capacity of

Fig. 6. Undrained stability of reconstructed embankment

198

ROY, SINGH, BHATTACHARYA, GHOSH & DASGUPTA ON redesign. Embankment reconstruction above 8 m height was allowed from May 2007 after these data were reviewed. Fill placement for the remaining height continued thereafter and paving work was completed by the beginning of June 2007. The reconstructed highway embankment was reopened for vehicular traffic by the middle of June 2007. Since then the MSE wall appears to be performing satisfactorily.

for the MSE wall, embankment and stabilizing berm at the end of construction was estimated at 1.22. The estimated yield acceleration is 0.04 g. For these assessments it was assumed that the consolidation of soils within the PVD -treated zone would be 75 per cent complete at end of construction and the effect of consolidation outside the PVD-treated zone was neglected. Although the computed factor of safety at end of construction is expected to be slightly smaller than the recommendation of IRC4, in longer term the factor of safety is expected to reach 1.38 upon completion of consolidation of soils within the PVDtreated zone. Also, since the yield acceleration is twothirds the design horizontal seismic coefficient, small permanent deformations are only expected as per the sliding block approach (see, e.g., Singh et al.8 for details) if the design earthquake were to occur in the short term.
9. RECONSTRUCTION

The PVDs (Colbonddrain CX1000) were installed in October 2006 at 1.2 m horizontal spacing in a square grid pattern using a mandrel of diamond shaped cross section with diagonals measuring 50 mm and 120 mm. The width of the PVD treated zone was 19 m measured west from the face of the failed MSE wall north of station 18.300. The width was 17 m south of station 18.300. The stabilizing berm was constructed between October 2006 and February 2007. The height of the upper bench was 3 m lower than the finished road grade and its width was 20 m. The maximum overall berm width measured from the base of the reconstructed MSE wall to the toe of the lower bench was 37 m. After removing the failed embankment to an elevation of about 5.75 m above the original ground surface, reconstruction work for the MSE wall was taken up in February 2007. Drilling, sampling and Standard Penetration Testing were carried out at BH1 and BH2 and VSTs were conducted at V4, V4A, and V5 in end April 2007 to check whether the gain in undrained shear strength due to accelerated consolidation of soft soils within the PVD treated zone was as assumed in MSE wall redesign. Data from these tests, presented in Fig. 7 together with the undrained shear strengths from nearby locations measured immediately after failure, indicate that the increased undrained shear strengths following accelerated consolidation of the soils within the PVD treated zone for the most part exceeded those assumed in the overall stability assessment for embankment

Fig. 7. Consolidation-related increase in undrained shear strength

10. MONITORING AND INSTRUMENTATION DURING RECONSTRUCTION

Three standpipe piezometers, P1, P2 and P3, and three settlement gages, S1, S2 and S3, were installed through the stabilizing berm. In addition, three standpipe piezometers, P4, P5 and P6, and three settlement gages, S4, S5, and S6, were installed within the footprint of MSE wall after the removal of the failed embankment. The locations of the instruments are shown in Fig. 8.

Fig. 8. Instrument locations

Data from these instruments, presented in Fig. 9, indicate that the settlements were continuing to develop and pore water pressures were still dissipating after about 4.5 months of preload placement (completed in the first

HIGHLIGHTS OF THE 178TH COUNCIL MEETING FAILURE OF MECHANICALLY STABILIZED EARTH WALL AT KM 18 OF NH 6 - A CASE HISTORY week of January 2007) and PVD installation (completed in the third week of October 2006). All the monitoring instruments at this site are serviceable at the time of preparation of this article. These instruments were systematically monitored over the entire duration of reconstruction and the monitoring activities continued until the expiry of two months after the beginning of operation of the highway overpass.
11. CONCLUSIONS

199

Case history pertaining to highway embankment failure at soft soil site has been presented. The site is located within the floodplains and backswamps of river Hooghly, a distributary of the Ganges underlain by soft silty clay and clayey silt. The km 18 site remains waterlogged throughout the year. The MSE wall construction at km 18 was initially undertaken over unimproved ground. The post failure investigation of the site indicated, on the other hand, that embankment failure was primarily caused by Overestimation of undrained shear strength, and Underestimation of driving force because of the use of smaller embankment heights and material unit weight in the overall stability assessment in the initial design This incident could be avoided by undertaking adequate geotechnical investigation at the design stage and appropriate coordination between the design office and construction activities. Data from pre- and post- failure geotechnical investigations from km 18 site indicate that the SHANSEP approach provides a reasonable guidance for estimating the undrained soil strengths of the soft foundation soils at this site. Two possible remedial measures for the redesign and reconstruction of the embankment with MSE wall, one were the use of piles and the other was the use of PVDs were found suitable. The second remedial measure was adopted mainly because of economics and convenience in contract administration. Before the reconstruction MSE wall at km 18 site PVDs were installed along the outer margins of the failed MSE wall followed by preload placement over the PVD treated area.

Fig. 9. Monitoring data

The MSE wall was reconstructed after adequate consolidation and strengthening of the soft soils within the PVD treated zone was achieved. From start to finish the reconstruction work at km 18 site spanned about six months. This site is now operational and appears to be performing satisfactorily.
12. POINTS FOR DISCUSSION

The paper presents a case history of failure of MSE wall coupled with settlement of a high embankment founded on weak soil underneath. The following points call for discussion from the professionals especially those associated with design and construction of MSE walls. 1. The investigation parameters, e.g., borehole spacing, depth of investigation, along and across the highway section involving high embankment construction. 2. Engineering properties of the subsoil to be assessed during the investigation process vis--vis the height of the embankment. 3. Appropriate techniques for subsoil investigation.

200

ROY, SINGH, BHATTACHARYA, GHOSH & DASGUPTA ON FAILURE OF MECHANICALLY STABILIZED EARTH WALL AT KM 18 OF NH 6 - A CASE HISTORY
3. Bureau of Indian Standards (BIS), Classification and Identification of Soils for General Engineering Purposes. IS 1498, BIS, New Delhi, 1970. IRC: 75-1979, Guidelines for the Design of High Embankments. The Indian Roads Congress, New Delhi, India, 1979. Interactive Software Designs Inc., XSTABL User manual, Moscow, ID, USA, 1994. Bishop, A.W., The Use of Slip Circle in Stability Analysis of Slopes. Gotechnique, 5 (1955), pp. 7-17. Ladd, C.C., Foott, R., Ishihara, K., Schlosser, F., and Poulos, H.G., Stress-deformation and Strength Characteristics. Proceedings, 9th International Conference on Soil Mechanics and Foundation Engineering, Tokyo, 1977, pp. 412-494. Singh, R., Roy, D., and Das, D., A Correlation for Permanent Earthquake Induced Deformation for Earth Embankments. Engineering Geology, 90 (2007), pp. 174185.

4. Considerations for design of high embankments and MSE walls with methodology. 5. Appropriate instrumentation scheme and monitoring schedule for coordination between design and construction activities. 6. To review the existing provisions in IRC 75, 1979.
ACKNOWLDGEMNETS

4.

5. 6. 7.

The authors would like to thank Mr. H. Fries Mr. P. Minhas, and Lt. Col. G.G.K. Nair of the project implementation team for technical help.
REFERENCES 1. Vaidyanadhan, R., and Ghosh, R.N., Quaternary of the East Coast of India. Current Science, Indian Academy of Sciences, 64 (1993), pp. 804-816. Gangopadhyay, C.R., and Das, S.C., The Design of the Approach Embankments of a Three-tier Flyover Structure at Joypur Beel, West Bengal. Journal of Indian National Society of Soil Mechanics and Foundation Engineering, 8 (1969), pp. 155-176.

8.

2.