The Mass Rapid Transit (MRT) rail network system is an integral part of Singapore’s public transport system. The MRT network involved construction of bored and cut and cover type tunnels. On the 20th of April 2004 at around 3:30 pm the temporary lateral support system of the cut and cover tunnel near the Nicoll Highway collapsed, resulting in the formation of a 100 ft. deep cave, which spread across six lanes of the Nicoll Highway (Figure 1). The collapse killed four people and injured three. The incident resulted in disruption of the gas, water and electric lines, which affected nearly 15000 people in the area. Two spans of a nearby bridge had to be demolished and reconstructed due to the damage in soil conditions the collapse had done in the nearby areas. One of the chief reasons for the failure of the temporary lateral support system was due to the overestimation of the undrained shear strength capacity of the soil.
Figure-1: Nicoll Highway Collapse, Source: COI 2005 (Permission Requested)
Keywords
Nicoll Highway, Deep Excavation, Cut and Cover Tunnels, Method A (Plaxis)
Background
Cut and Cover Tunnel: Construction of cut and cover tunnel involves building the tunnel structure inside an excavation. After the completion of the tunnel structure the excavation is covered with backfill material. Cut and cover type tunnels are typically constructed for depths of up to 30-40 feet. There are two approaches to the construction of a cut and cover tunnel (FHA Handbook 2009, pp 5-1-5-12). a) The Bottom-Up Method. b) The Top-Down Method.
I.Bottom-Up method Bottom-up method involves excavation of a trench from the surface. The sides of the excavation can be sloped or vertical. The tunnel is then constructed in the trench and after the completion of the tunnel the trench is backfilled. Generally in urban areas due to limited availability of space the tunnel is often constructed with a vertical excavation supported by excavation support system. Excavation support system can be temporary or permanent. Some of the temporary excavation systems are sheet pile walls with multi levels of bracing, soldier piles with lagging walls, tieback support systems. Permanent excavation support systems form the part of the final tunnel structure. Some of the permanent excavation support systems are slurry walls, tangent pile wall support and soldier pile tremie concrete. II.The Top-Down method Top-down method involves the construction of the tunnel walls, which are made using slurry walls, secant pile walls. After the construction of the tunnel walls the roof slab is constructed on the ground and connected with the tunnel walls. The roof slab is then covered and the surface of the ground can be used. Rest of the tunnel is excavated below the protection of the roof slab and the tunnel walls. After the excavation the rest of the tunnel can be finished and the floor slab can be constructed. This method is challenging to construct but the ground surface above the tunnel can be restored earlier. Bottom-Up method was used to construct the part of the cut and cover tunnel near Nicoll highway. Diaphragm walls supported the sides of the excavation and multi-levels of bracing were constructed as the excavation progressed.
Design and Construction of the Tunnel
The site soil conditions were assessed by a general site investigation conducted few years before the contract was awarded. The investigation included 14 boreholes and cone penetration tests along with collection of soil samples and testing them in laboratories. After the contract was awarded an extensive testing was conducted which included 72 boreholes and cone penetration tests to determine the alignment of the tunnel and get data for the purpose of design (COI 2005, pp 16-35). The tunnels were constructed on reclaimed lands. Reclamation was done in two stages. Northern part of the site was reclaimed over 50 years ago and the southern part was reclaimed about 20 years ago. Therefore the top soil of the collapse site comprised of fill material. Underneath the top layer there were layers of soft clays (marine clay and estuarine clay deposits) to a depth of 37-40 meters. Beneath these layers were layers of sands, silts, clays and an old alluvium formation. Ground water table was generally about 2 meters below the surface. For design purposes diaphragm walls were divided into 40 different wall sections. Wall sections were designed based on the worst soil parameters obtained from the nearest borehole. The temporary retaining wall system consisting of diaphragm walls and struts was modeled, analyzed and designed using Plaxis, which is a widely used geotechnical modeling software based on finite element method. Soil layers were modeled using Mohr-Coulomb soil model with effective stress strength parameters. A load factor of 1.2 was considered for obtaining the design forces on the diaphragm. Diaphragm wall designs were optimized based on the wall movement criteria, with maximum allowable movement being 200 mm at any depth of the wall and 40 mm at the diaphragm wall toe. Wall sections of type M2, M3 failed during the collapse. Typical cross-section of the failed tunnel section is shown in figure 3. The diaphragm walls were 0.8-1 m thick and were about 40-45 m deep. Walls extended about 1-3 m into the stiffer strata. Width of the tunnel was about 17-20 m and width of each diaphragm wall panel was 6 m. There were two layers of interlocking jet grout piles. The upper layer of the jet grout pile was 1.5 m thick and was temporary and the lower layer of the jet grout pile was 2.5 m thick and formed the base of the tunnel. Jet grout pile layers were built to minimize the deflection of the walls while the tunnel was being excavated. Bored piles were constructed to support the rail boxes. Excavation was supported by a system of steel king post and 10 levels of struts placed at 4m center to center. As the excavation progressed the struts were constructed and before the construction of the 10th level of strut the temporary layer of the jet gout pile was removed. The entire process of tunnel construction was monitored by thousands of geotechnical instruments including settlement markers, inclinometers to monitor the soil and wall deflections, vibrating wire piezometers, strain gauges and load cells. The instruments installed in the failed part of the tunnel section provided data, which helped to understand the reasons for the collapse. Figure 2 shows the part of the section collapsed. The inclinometers, strain gauges and load cells were installed on the strut “S-335” (Red).
Figure-2: Collapsed portion of the tunnel, Source: Tunnels and Tunneling Magazine (Permission Requested)
Figure-3: Cross-section of Tunnel, Source: Tunnels and Tunneling Magazine (Permission Requested)
Events Leading Up to Collapse
According to the Committee of Inquiry report “The Collapse did not develop suddenly. A chain of events preceded the collapse that in retrospect appeared to be a warning of the problems that were developing”. There were many warnings of approaching collapse but most of the warnings were either not taken seriously or ignored. Some of the incidents that happened prior to the collapse shed some light on the critical error in the design methodology that is the use of effective stress approach for the design of the diaphragm walls of the temporary retaining wall systems.
One of the major incidents that happened prior to the collapse was in the tunnel launch shaft area in August 2003. Tunnel launch shaft was located at the eastern end of the cut and cover tunnels. During the excavation of the tunnel launch shaft at about the 7th level of struts severe deflections of the diaphragm walls was observed which exceeded the design limit. Excessive ground settlement in the order of 400 mm was observed in a stadium nearby the south diaphragm wall (COI 2005, pp 36-54). Vertical cracks in some of the diaphragm wall panels were observed. Further excavation of the tunnel launch shaft was stopped immediately.
A back analysis was performed for the diaphragm walls based on effective stress approach and it was determined that the walls were not capable of withstanding the load if the excavation had proceeded. So the contractor constructed a jet grout pile layer and temporary walls to support the deflected wall and added additional struts to support the excavation before the excavation was resumed.
The critical error of using the effective stress approach in place of total stress approach overestimates the undrained shear strength of the marine clays which results in underestimation of the deflection, bending moment in the walls. This explanation was pointed out by the engineering advisory panel of the owners. They strongly suggested the contractors to reanalyze the entire wall sections using the total stress approach as it was a very unsafe basis for design. The contractor declined to reanalyze claiming that their design was sound and the fact that the excavation in other areas based on the effective stress approach behaved as designed. They eventually agreed to do a back analysis based on the total stress approach for the tunnel shaft area.
Similar problems arose in many of the other wall sections, excessive deflections of the wall and cracks in the diaphragm panels. There were excessive ground settlements in many properties near the alignment of the tunnels. These problems were fixed by the contractor in a similar fashion based on a back analysis with the effective stress method. The contractors were reluctant to use the total stress approach for back analysis as they felt that the method gave large wall deflections, which would result in thicker walls. They also justified the use of effective stress approach based on the fact that the analysis provided similar results as the actual deflection of the walls. However this was partially true as for smaller depths the effective stress method would give close displacement results but as the excavation in the marine clay progresses the deflections calculated become highly un-conservative and wouldn’t match the actual deflection and so was still a very unsafe basis for design. Although this was not very clear at that point of time so the owners agreed to the further excavation of the various tunnel sections.
The failed wall sections M2 and M3 near the Nicoll highway displayed similar problems to the other wall sections like excessive deflections of the wall, which was way above the design limits as early as November of 2003. The issues were resolved in a similar fashion by performing back analyses based on the effective stress method followed by appropriate remediation measures to support the excavation and then continuing on with the excavation. The owners agreed to the approach of using the effective stress methods for back analysis and remediation based on that analysis and suggested that the contractor should monitor the data from the inclinometers at all levels and adopt contingency measures if the values exceeded the design values. The critical error of using effective stress method for analysis and subsequent back analysis became more evident as the excavation progressed deeper. As the excavation of different sections of the tunnel were carried out in parallel, the excavation didn’t reach to such levels were the diaphragm walls would have failed. But the tunnel section near the Nicoll highway had to be excavated quicker than other sections as the construction and excavation work hindered with the traffic flow on the highway.
As the excavation of the M3 type walls progressed to about the 6th level of strut the deflection of the walls exceeded the design limit in February of 2004. The section was back analyzed using effective stress approach and the design limits were revised. The excavation progressed with the new revised deflection limits. By the end of March of 2004 the deflection of the walls had exceeded the revised deflection limits. A second back analysis was done based on the same effective stress method and the design deflection limits were further revised and remediation measures were taken to keep the excavation open. The owners accepted the second back analysis and permitted the further excavation on 3rd April 2004. Between 3rd April and the 20th April there were periods of time when the inclinometer readings were not monitored by the contractor.
In the meanwhile the excavation progressed to about the 10th level of struts. The temporary layer of jet grout piles below the 10th level which was constructed to add to the stability of the excavation was now removed to finish the excavation of the tunnel. Two more temporary struts were added to stabilize the excavation. By 19th April the excavation of the tunnel was completed and the last bit of the tunnel section connecting it with the other section of the tunnel was cut out and the tunnels were connected. On the day of the collapse the 20th of April at around 8:00 am the workers at the site heard sounds from the multi-level strut system. The strange sounds were investigated by the senior engineers of the contractor and they found that a lot of the waler-strut connections had yielded. They immediately instructed everyone to leave the worksite. By noon engineers from the owner’s side were also present at the worksite and they along with the contractor’s engineers decided to pour concrete at the 9th level of the strut to stabilize the excavation and prevent it from caving in. By 3:30 pm the temporary system gave away and the tunnel had caved in.
Reasons for Collapse
There are many theories for the collapse of the tunnel near Nicoll Highway. Some of the theories are presented in this section.
A.Official Reasons for Collapse:
1.The use “Method A (Effective Stress Method)” for the design of the diaphragm walls. The committee of inquiry found that the main reason for the collapse of the temporary retaining wall system was due to the under-design of the diaphragm wall using Method ‘A’ of Plaxis, which is an effective stress method (COI 2005, pp 1-16). This method is inappropriate to model deep excavations in soft marine clays. Rapid excavations in slightly overconsolidated soft marine clays does not give sufficient time for the excess pore water pressure in the clays to dissipate. Therefore the geotechnical analysis should be performed based on undrained conditions. Total stress method is usually applied for analysis in such soil conditions (A.M. Purzin et al. 2010, pp 160-161). The use of the effective stress method (Method ‘A’) resulted in overestimation of the undrained shear strength, which resulted in underestimation of the diaphragm wall bending moment and wall deflections. The use of Method ‘A’ resulted in underestimation of the bending moments and deflection by about 50%.
2.Under-design of the waler-strut connection. Another chief reason for the failure was the under design of the waler-strut connection. Designers misinterpreted the stiff bearing length for the C-channels as per BS 5950 to 400 mm instead of using 65 mm. Waler connections with C-channels were designed with the effective length factor of 0.7, where the end conditions were unrestrained and a factor of 1.2 should have been used. This resulted in axial design capacity that was about 70% of the assumed design load for the connection (COI 2005, pp 1-16). Also in some locations the splays in the waler strut connection had been omitted during construction figure 4 (NCE, 2005). These design and construction errors resulted in the failure of the 9th level strut-waler system. The under designed diaphragm wall could not resist the redistributed loads as the 9th level strut-waler system failed resulting in the collapse of the tunnel.
1.Forced sway failure mode According to the contractors a relative vertical displacement of the wall with respect to the king posts triggered the collapse of the tunnel (NCE, 2005). This sudden drop in height altered the angle at which the struts were connected to the walers causing sway type failure. Finite element models of the waler-strut connections corroborated with the fact that C-channel connection was susceptible to sway type failure. Forced sway failure mechanism is associated with brittle failure of the connection and was the reason why the temporary wall retaining system failed in about 6 hrs. If this failure had not occurred the failure could have taken days or weeks giving the contractor sufficient time to stabilize the excavation.
2.Lack of toe in depth of the diaphragm wall into competent strata In the investigations following the collapse it was found that certain parts of the tunnel went through a buried valley of old alluvium under the layers of marine clays. Investigation revealed that the upper portion of the old alluvium valley consisted of sands, silts and clay with organic material with very low SPT (Standard Penetration Test) N-values (about N=30). This fact was unknown at the time of design as there were very few test boreholes in the area. So in certain sections the toe of the diaphragm wall was not embedded sufficiently into stiffer strata (K. Ishihara 2011, pp 35-50).
Lessons Learned
The main reason for the failure of the tunnel near the Nicoll Highway was due to the under-design of the diaphragm walls. There are many important lessons to be learned from the collapse. Although this flaw in design was known before and pointed out, the full impact of the flaw was not properly understood at the time of the design. So it is very essential that specialists with appropriate experience should design works of public importance. Design documents should be subjected to better reviews and checks. Precaution must be taken while using new or unfamiliar technologies and the limitations of such technologies should be properly understood.
Remediation
Nicoll highway was closed to traffic for almost 8 months to stabilize the collapse site. The collapse disturbed the soil conditions in the nearby area so it was decided to change the alignment of the tunnel to avoid the collapsed area. The new alignment ran 100 m south of the original alignment. A totally new approach of design and construction was adopted for the new alignment. The tunnels were constructed using the tunnel-boring machine; the section of the tunnel which was near the stations were constructed using the top-down approach of cut and cover tunneling. Better quality control was affected by strict monitoring regime and use of automatic data loggers to continuously record and review data (LTA Safety Magazine 2005, pp 9-14).
Bibliography
A.M. Purzin et al. (2010). “Braced Excavation Collapse: Nicoll Highway, Singapore”. Geomechanics of Failures. pp 151-181
The author presents the various theories for the collapse of the braced excavation.
A.J. Whittle and Gonzalo Corral (2010). “Re-analysis of Deep Excavation Collapse Using a Generalized Effective Stress Soil Model”. Earth Retention Conference 2010 – ASCE, 1-4 August 2010, Bellevue, WA.
The authors analyzed the excavation based on soil parameters from high quality consolidation test and undrained triaxial shear test. The wall deflections and strut loads calculated reasonably matched the measured data.
A.J. Whittle and R.V. Davies (2006). “Nicoll Highway Collapse: Evaluation of Geotechnical Factors Affecting Design of Excavation Support System”. International Conference on Deep Excavations 28-30 June 2006, Singapore
Authors evaluate the soil parameters and find that the reclaimed marine clay was under consolidated which resulted in low undrained shear strength of the clay.
C.F. Leung and S.A. Tan (2007). “Successes and Failures of Instrumentation Programs in Major Construction Projects in Singapore”. Seventh International Symposium on Field Measurements in Geomechanics, ASCE
The author examines the role of instrumentation in select major construction projects in Singapore. The author points out the poor attention to the instrumentation data, from the temporary braced supports led to the collapse of the Nicoll highway.
COI (2005). “Report of the Committee of Inquiry into the incident at the MRT Circle Line worksite that led to collapse of Nicoll Highway on 20th April 2004”. Ministry of Manpower, Singapore.
The Committee of Inquiry report identifies the key errors in the design resulting in the failure of the temporary lateral earth support system. The errors were under-design of the diaphragm wall and under-design of the waler connections in the strutting system.
The website gives a detailed overview of the cut and cover type tunnel construction and the types of temporary supports used during excavation.
Gouw Tjie-Liong (2012). “Deep Excavation Failures Can They Be Prevented”. International Symposium on Sustainable Geosynthetics and Green Technology for Climate Change. 20-21 June 2012, Bangkok
The Author discusses the technical and non-technical factors behind the collapse of tunnel near Nicoll Highway.
K. Ishihara (2011). “Collapse of Braced Excavation in Singapore”. International Symposium on Backward Problems in geotechnical Engineering, TC302-Osaka 2011
The author discusses about the lack of toe-in depth of the diaphragm wall into stiff competent strata as one of the reasons for the incident.
K.Y. Yong and S.L Lee (2007). “Collapse of Nicoll Highway- A Global Failure at the Curved Section of a Cut and Cover Tunnel Construction”. Journal of the South East Asian Geotechnical Society. pp 139-153
The paper presents the shortcomings in design and construction, and the influence of curved alignment of the tunnel near Nicoll Highway.
New Civil Engineer (2005). “Design and construction failures caused Singapore tunnel”. New Civil Engineer May 2005
The article presents the causes of the failure and presents the client’s view along with the contractor’s view on what went wrong.
New Civil Engineer (2005). “Contractor Nishimatsu Blamed For Nicoll Highway Collapse”. New Civil Engineer May 2005
The article discusses about the legal consequences of the failure.
Safety News (2005). “Rebuilding Nicoll Highway”. Land Transport Authority August 2005
The article presents details on the remediation of the collapsed section of the Nicoll Highway.
Singapore - April 20th, 2004
Shuvrajit Ghosh, M.Eng (Structural), The Pennsylvania State University (Fall 2014)
Table of Contents
Introduction
The Mass Rapid Transit (MRT) rail network system is an integral part of Singapore’s public transport system. The MRT network involved construction of bored and cut and cover type tunnels. On the 20th of April 2004 at around 3:30 pm the temporary lateral support system of the cut and cover tunnel near the Nicoll Highway collapsed, resulting in the formation of a 100 ft. deep cave, which spread across six lanes of the Nicoll Highway (Figure 1). The collapse killed four people and injured three. The incident resulted in disruption of the gas, water and electric lines, which affected nearly 15000 people in the area. Two spans of a nearby bridge had to be demolished and reconstructed due to the damage in soil conditions the collapse had done in the nearby areas. One of the chief reasons for the failure of the temporary lateral support system was due to the overestimation of the undrained shear strength capacity of the soil.
Keywords
Nicoll Highway, Deep Excavation, Cut and Cover Tunnels, Method A (Plaxis)
Background
Cut and Cover Tunnel:
Construction of cut and cover tunnel involves building the tunnel structure inside an excavation. After the completion of the tunnel structure the excavation is covered with backfill material. Cut and cover type tunnels are typically constructed for depths of up to 30-40 feet. There are two approaches to the construction of a cut and cover tunnel (FHA Handbook 2009, pp 5-1-5-12).
a) The Bottom-Up Method.
b) The Top-Down Method.
I. Bottom-Up method
Bottom-up method involves excavation of a trench from the surface. The sides of the excavation can be sloped or vertical. The tunnel is then constructed in the trench and after the completion of the tunnel the trench is backfilled. Generally in urban areas due to limited availability of space the tunnel is often constructed with a vertical excavation supported by excavation support system. Excavation support system can be temporary or permanent. Some of the temporary excavation systems are sheet pile walls with multi levels of bracing, soldier piles with lagging walls, tieback support systems. Permanent excavation support systems form the part of the final tunnel structure. Some of the permanent excavation support systems are slurry walls, tangent pile wall support and soldier pile tremie concrete.
II. The Top-Down method
Top-down method involves the construction of the tunnel walls, which are made using slurry walls, secant pile walls. After the construction of the tunnel walls the roof slab is constructed on the ground and connected with the tunnel walls. The roof slab is then covered and the surface of the ground can be used. Rest of the tunnel is excavated below the protection of the roof slab and the tunnel walls. After the excavation the rest of the tunnel can be finished and the floor slab can be constructed. This method is challenging to construct but the ground surface above the tunnel can be restored earlier.
Bottom-Up method was used to construct the part of the cut and cover tunnel near Nicoll highway. Diaphragm walls supported the sides of the excavation and multi-levels of bracing were constructed as the excavation progressed.
Design and Construction of the Tunnel
The site soil conditions were assessed by a general site investigation conducted few years before the contract was awarded. The investigation included 14 boreholes and cone penetration tests along with collection of soil samples and testing them in laboratories. After the contract was awarded an extensive testing was conducted which included 72 boreholes and cone penetration tests to determine the alignment of the tunnel and get data for the purpose of design (COI 2005, pp 16-35).
The tunnels were constructed on reclaimed lands. Reclamation was done in two stages. Northern part of the site was reclaimed over 50 years ago and the southern part was reclaimed about 20 years ago. Therefore the top soil of the collapse site comprised of fill material. Underneath the top layer there were layers of soft clays (marine clay and estuarine clay deposits) to a depth of 37-40 meters. Beneath these layers were layers of sands, silts, clays and an old alluvium formation. Ground water table was generally about 2 meters below the surface.
For design purposes diaphragm walls were divided into 40 different wall sections. Wall sections were designed based on the worst soil parameters obtained from the nearest borehole. The temporary retaining wall system consisting of diaphragm walls and struts was modeled, analyzed and designed using Plaxis, which is a widely used geotechnical modeling software based on finite element method. Soil layers were modeled using Mohr-Coulomb soil model with effective stress strength parameters. A load factor of 1.2 was considered for obtaining the design forces on the diaphragm. Diaphragm wall designs were optimized based on the wall movement criteria, with maximum allowable movement being 200 mm at any depth of the wall and 40 mm at the diaphragm wall toe.
Wall sections of type M2, M3 failed during the collapse. Typical cross-section of the failed tunnel section is shown in figure 3. The diaphragm walls were 0.8-1 m thick and were about 40-45 m deep. Walls extended about 1-3 m into the stiffer strata. Width of the tunnel was about 17-20 m and width of each diaphragm wall panel was 6 m. There were two layers of interlocking jet grout piles. The upper layer of the jet grout pile was 1.5 m thick and was temporary and the lower layer of the jet grout pile was 2.5 m thick and formed the base of the tunnel. Jet grout pile layers were built to minimize the deflection of the walls while the tunnel was being excavated. Bored piles were constructed to support the rail boxes. Excavation was supported by a system of steel king post and 10 levels of struts placed at 4m center to center. As the excavation progressed the struts were constructed and before the construction of the 10th level of strut the temporary layer of the jet gout pile was removed.
The entire process of tunnel construction was monitored by thousands of geotechnical instruments including settlement markers, inclinometers to monitor the soil and wall deflections, vibrating wire piezometers, strain gauges and load cells. The instruments installed in the failed part of the tunnel section provided data, which helped to understand the reasons for the collapse. Figure 2 shows the part of the section collapsed. The inclinometers, strain gauges and load cells were installed on the strut “S-335” (Red).
Events Leading Up to Collapse
According to the Committee of Inquiry report “The Collapse did not develop suddenly. A chain of events preceded the collapse that in retrospect appeared to be a warning of the problems that were developing”. There were many warnings of approaching collapse but most of the warnings were either not taken seriously or ignored. Some of the incidents that happened prior to the collapse shed some light on the critical error in the design methodology that is the use of effective stress approach for the design of the diaphragm walls of the temporary retaining wall systems.
One of the major incidents that happened prior to the collapse was in the tunnel launch shaft area in August 2003. Tunnel launch shaft was located at the eastern end of the cut and cover tunnels. During the excavation of the tunnel launch shaft at about the 7th level of struts severe deflections of the diaphragm walls was observed which exceeded the design limit. Excessive ground settlement in the order of 400 mm was observed in a stadium nearby the south diaphragm wall (COI 2005, pp 36-54). Vertical cracks in some of the diaphragm wall panels were observed. Further excavation of the tunnel launch shaft was stopped immediately.
A back analysis was performed for the diaphragm walls based on effective stress approach and it was determined that the walls were not capable of withstanding the load if the excavation had proceeded. So the contractor constructed a jet grout pile layer and temporary walls to support the deflected wall and added additional struts to support the excavation before the excavation was resumed.
The critical error of using the effective stress approach in place of total stress approach overestimates the undrained shear strength of the marine clays which results in underestimation of the deflection, bending moment in the walls. This explanation was pointed out by the engineering advisory panel of the owners. They strongly suggested the contractors to reanalyze the entire wall sections using the total stress approach as it was a very unsafe basis for design. The contractor declined to reanalyze claiming that their design was sound and the fact that the excavation in other areas based on the effective stress approach behaved as designed. They eventually agreed to do a back analysis based on the total stress approach for the tunnel shaft area.
Similar problems arose in many of the other wall sections, excessive deflections of the wall and cracks in the diaphragm panels. There were excessive ground settlements in many properties near the alignment of the tunnels. These problems were fixed by the contractor in a similar fashion based on a back analysis with the effective stress method. The contractors were reluctant to use the total stress approach for back analysis as they felt that the method gave large wall deflections, which would result in thicker walls. They also justified the use of effective stress approach based on the fact that the analysis provided similar results as the actual deflection of the walls. However this was partially true as for smaller depths the effective stress method would give close displacement results but as the excavation in the marine clay progresses the deflections calculated become highly un-conservative and wouldn’t match the actual deflection and so was still a very unsafe basis for design. Although this was not very clear at that point of time so the owners agreed to the further excavation of the various tunnel sections.
The failed wall sections M2 and M3 near the Nicoll highway displayed similar problems to the other wall sections like excessive deflections of the wall, which was way above the design limits as early as November of 2003. The issues were resolved in a similar fashion by performing back analyses based on the effective stress method followed by appropriate remediation measures to support the excavation and then continuing on with the excavation. The owners agreed to the approach of using the effective stress methods for back analysis and remediation based on that analysis and suggested that the contractor should monitor the data from the inclinometers at all levels and adopt contingency measures if the values exceeded the design values. The critical error of using effective stress method for analysis and subsequent back analysis became more evident as the excavation progressed deeper. As the excavation of different sections of the tunnel were carried out in parallel, the excavation didn’t reach to such levels were the diaphragm walls would have failed. But the tunnel section near the Nicoll highway had to be excavated quicker than other sections as the construction and excavation work hindered with the traffic flow on the highway.
As the excavation of the M3 type walls progressed to about the 6th level of strut the deflection of the walls exceeded the design limit in February of 2004. The section was back analyzed using effective stress approach and the design limits were revised. The excavation progressed with the new revised deflection limits. By the end of March of 2004 the deflection of the walls had exceeded the revised deflection limits. A second back analysis was done based on the same effective stress method and the design deflection limits were further revised and remediation measures were taken to keep the excavation open. The owners accepted the second back analysis and permitted the further excavation on 3rd April 2004. Between 3rd April and the 20th April there were periods of time when the inclinometer readings were not monitored by the contractor.
In the meanwhile the excavation progressed to about the 10th level of struts. The temporary layer of jet grout piles below the 10th level which was constructed to add to the stability of the excavation was now removed to finish the excavation of the tunnel. Two more temporary struts were added to stabilize the excavation. By 19th April the excavation of the tunnel was completed and the last bit of the tunnel section connecting it with the other section of the tunnel was cut out and the tunnels were connected. On the day of the collapse the 20th of April at around 8:00 am the workers at the site heard sounds from the multi-level strut system. The strange sounds were investigated by the senior engineers of the contractor and they found that a lot of the waler-strut connections had yielded. They immediately instructed everyone to leave the worksite. By noon engineers from the owner’s side were also present at the worksite and they along with the contractor’s engineers decided to pour concrete at the 9th level of the strut to stabilize the excavation and prevent it from caving in. By 3:30 pm the temporary system gave away and the tunnel had caved in.
Reasons for Collapse
There are many theories for the collapse of the tunnel near Nicoll Highway. Some of the theories are presented in this section.
A. Official Reasons for Collapse:
1. The use “Method A (Effective Stress Method)” for the design of the diaphragm walls.
The committee of inquiry found that the main reason for the collapse of the temporary retaining wall system was due to the under-design of the diaphragm wall using Method ‘A’ of Plaxis, which is an effective stress method (COI 2005, pp 1-16). This method is inappropriate to model deep excavations in soft marine clays. Rapid excavations in slightly overconsolidated soft marine clays does not give sufficient time for the excess pore water pressure in the clays to dissipate. Therefore the geotechnical analysis should be performed based on undrained conditions. Total stress method is usually applied for analysis in such soil conditions (A.M. Purzin et al. 2010, pp 160-161). The use of the effective stress method (Method ‘A’) resulted in overestimation of the undrained shear strength, which resulted in underestimation of the diaphragm wall bending moment and wall deflections. The use of Method ‘A’ resulted in underestimation of the bending moments and deflection by about 50%.
2. Under-design of the waler-strut connection.
Another chief reason for the failure was the under design of the waler-strut connection. Designers misinterpreted the stiff bearing length for the C-channels as per BS 5950 to 400 mm instead of using 65 mm. Waler connections with C-channels were designed with the effective length factor of 0.7, where the end conditions were unrestrained and a factor of 1.2 should have been used. This resulted in axial design capacity that was about 70% of the assumed design load for the connection (COI 2005, pp 1-16). Also in some locations the splays in the waler strut connection had been omitted during construction figure 4 (NCE, 2005). These design and construction errors resulted in the failure of the 9th level strut-waler system. The under designed diaphragm wall could not resist the redistributed loads as the 9th level strut-waler system failed resulting in the collapse of the tunnel.
B. Other Theories for the Collapse:
1. Forced sway failure mode
According to the contractors a relative vertical displacement of the wall with respect to the king posts triggered the collapse of the tunnel (NCE, 2005). This sudden drop in height altered the angle at which the struts were connected to the walers causing sway type failure. Finite element models of the waler-strut connections corroborated with the fact that C-channel connection was susceptible to sway type failure. Forced sway failure mechanism is associated with brittle failure of the connection and was the reason why the temporary wall retaining system failed in about 6 hrs. If this failure had not occurred the failure could have taken days or weeks giving the contractor sufficient time to stabilize the excavation.
2. Lack of toe in depth of the diaphragm wall into competent strata
In the investigations following the collapse it was found that certain parts of the tunnel went through a buried valley of old alluvium under the layers of marine clays. Investigation revealed that the upper portion of the old alluvium valley consisted of sands, silts and clay with organic material with very low SPT (Standard Penetration Test) N-values (about N=30). This fact was unknown at the time of design as there were very few test boreholes in the area. So in certain sections the toe of the diaphragm wall was not embedded sufficiently into stiffer strata (K. Ishihara 2011, pp 35-50).
Lessons Learned
The main reason for the failure of the tunnel near the Nicoll Highway was due to the under-design of the diaphragm walls. There are many important lessons to be learned from the collapse. Although this flaw in design was known before and pointed out, the full impact of the flaw was not properly understood at the time of the design. So it is very essential that specialists with appropriate experience should design works of public importance. Design documents should be subjected to better reviews and checks. Precaution must be taken while using new or unfamiliar technologies and the limitations of such technologies should be properly understood.
Remediation
Nicoll highway was closed to traffic for almost 8 months to stabilize the collapse site. The collapse disturbed the soil conditions in the nearby area so it was decided to change the alignment of the tunnel to avoid the collapsed area. The new alignment ran 100 m south of the original alignment. A totally new approach of design and construction was adopted for the new alignment. The tunnels were constructed using the tunnel-boring machine; the section of the tunnel which was near the stations were constructed using the top-down approach of cut and cover tunneling. Better quality control was affected by strict monitoring regime and use of automatic data loggers to continuously record and review data (LTA Safety Magazine 2005, pp 9-14).
Bibliography
A.M. Purzin et al. (2010). “Braced Excavation Collapse: Nicoll Highway, Singapore”. Geomechanics of Failures. pp 151-181
A.J. Whittle and Gonzalo Corral (2010). “Re-analysis of Deep Excavation Collapse Using a Generalized Effective Stress Soil Model”. Earth Retention Conference 2010 – ASCE, 1-4 August 2010, Bellevue, WA.
A.J. Whittle and R.V. Davies (2006). “Nicoll Highway Collapse: Evaluation of Geotechnical Factors Affecting Design of Excavation Support System”. International Conference on Deep Excavations 28-30 June 2006, Singapore
C.F. Leung and S.A. Tan (2007). “Successes and Failures of Instrumentation Programs in Major Construction Projects in Singapore”. Seventh International Symposium on Field Measurements in Geomechanics, ASCE
COI (2005). “Report of the Committee of Inquiry into the incident at the MRT Circle Line worksite that led to collapse of Nicoll Highway on 20th April 2004”. Ministry of Manpower, Singapore.
Federal Highway Administration. “Technical Manual for Design and Construction of Road Tunnels - Civil Elements”. Website of Federal Highway Administration. http://www.fhwa.dot.gov/bridge/tunnel/pubs/nhi09010/05.cfm. Chapter 5
Gouw Tjie-Liong (2012). “Deep Excavation Failures Can They Be Prevented”. International Symposium on Sustainable Geosynthetics and Green Technology for Climate Change. 20-21 June 2012, Bangkok
K. Ishihara (2011). “Collapse of Braced Excavation in Singapore”. International Symposium on Backward Problems in geotechnical Engineering, TC302-Osaka 2011
K.Y. Yong and S.L Lee (2007). “Collapse of Nicoll Highway- A Global Failure at the Curved Section of a Cut and Cover Tunnel Construction”. Journal of the South East Asian Geotechnical Society. pp 139-153
New Civil Engineer (2005). “Design and construction failures caused Singapore tunnel”. New Civil Engineer May 2005
New Civil Engineer (2005). “Contractor Nishimatsu Blamed For Nicoll Highway Collapse”. New Civil Engineer May 2005
Safety News (2005). “Rebuilding Nicoll Highway”. Land Transport Authority August 2005
Singapore Infopedia. “ Nicoll Highway Collapse “. Singapore government website.
http://eresources.nlb.gov.sg/infopedia/articles/SIP_430_2004-12-17.html
Wikipedia. “Nicoll Highway Collapse”. http://en.wikipedia.org/wiki/Nicoll_Highway_collapse