Seismic safety retrofit of a major bridge – A Canadian study

U. D. Atukorala & H. Puebla Golder Associates Ltd., Burnaby, , Canada

ABSTRACT: Key aspects of a displacement-based design that was adopted to retrofit seismic deficiencies of a major bridge in , Canada, are presented. The displacements anticipated as a result of the 475-year seismic loading were computed using a state-of-the art stress-strain model developed at the University of British Columbia (Canada) that models the observed post-liquefaction behaviour of soil. This stress-strain model was incorporated into the public domain finite difference program FLAC2D for the analysis of deformations. The results were used to arrive at an optimum ground improvement scheme at selected critical locations along the bridge profile rather than at each and every pier location. Where ground improvement using conventional methods such as vibro-compaction and gravel compaction piles was not feasible due to access and headroom constraints, pore pressure dissipation measures were considered using seismic drains. The paper presents some of the results of the ground response analyses and discusses the ground improvement program.

1 INTRODUCTION

Vancouver and the densely populated Lower Mainland region of British Columbia, Canada, are in an area of high seismic risk associated with a major subduction zone off the Canadian west coast. The dynamic geological setting makes this region one of the most seismically active regions of Canada. Although Vancouver has not yet experienced a major earthquake, a number of large earthquakes have occurred in the region that have been felt in Vancouver. The 1946 Campbell River earthquake is the largest magnitude earthquake that has affected the Lower Mainland, and it is estimated that the area was subjected to a peak bedrock acceleration of about 0.06 g. While no significant damage has been reported, some of the newspapers had reported that the ground near the Vancouver International Airport rolled in waves (Byrne & Anderson, 1987). The earthquake environment in the region is similar to that of the east coast of Japan, the south coast of Alaska, and most of the west coast of Central and South America (Rogers, 1998). Crustal and subcrustal earthquakes of magnitudes varying from M6 to 7 have been recorded over the last century within the North American and the subducting Juan de Fuca plates that extend beneath Vancouver. Geological evidence also suggests that there is the potential for a much larger inter-plate earthquake of magnitude M8+ occurring within the off-shore Cascadia subduction zone. A major earthquake associated with either of these sources could have devastating effects in Vancouver and the Lower Mainland Region of British Columbia. The infrastructure of the Lower Mainland Region comprises a number of bridges that serve as major transportation links, connecting Vancouver to its neighbouring municipalities, and form key elements of the designated Disaster Response Routes. Most of these bridges were constructed in the 1950’s when earthquake design considerations were not as stringent as they are in the current bridge design codes, and at a time when phenomena such as “soil liquefaction” were not well understood. This paper describes the strategy followed in the seismic retrofit of the foundations of Second Narrows Bridge in Vancouver, Canada. The design phase was completed in May 2001, and the foundation retrofit strategy is currently being implemented (October 2002).

Paper Number 098 2 SEISMIC DESIGN CRITERIA

The BC Ministry of Transportation and Highways (May 2000) instituted a seismic risk reduction program for all highway bridges in British Columbia to minimize public risk and preserve important transportation routes for use after a major earthquake. The program implements stringent seismic design standards for planned new bridges and “seismic retrofit” measures to improve the seismic resistance/performance of existing bridges. The key objectives of the program are: · Minimize risk of collapse of the bridge or unacceptable damage when subjected to earthquake shaking that corresponds to a 10% chance of being exceeded in 50 years or a return period of 475 years; · Preserve important highway routes for disaster response and economic recovery after a major earthquake; and · Reduce damage and minimize loss of life and/or injury after a major earthquake. For existing bridges, several retrofit levels are considered with corresponding performance requirements/expectations:

Table 1: Seismic Performance Criteria Retrofit Level Seismic Performance Criteria Service Level Damage Level Functional Retrofit Immediate – Full access to normal traffic is Minimal available almost immediately following the earthquake. Safety Retrofit – I Limited – Limited access is available within Repairable hours following the earthquake. Full access to normal traffic is restorable within days. Safety Retrofit – II Significantly Limited – It is expected, but not Significant guaranteed, that limited access to light emergency traffic is possible within days following the earthquake. No public access is possible until repairs are completed. Superstructure Possible complete loss of service. Possible Collapse But Retrofit Significantly Reduced Risk A significant area of the Lower Mainland is underlain by recent unconsolidated sediments deposited by several rivers and creeks in the region. Several important bridges are located in areas comprising unconsolidated sediments that are in loose to compact state. The Second Narrows Bridge, which spans over the and connects the City of Vancouver with the District of (see Figure 1), is located in an area with unconsolidated sediments beneath the mouth of Seymour River.

3 SEISMIC RETROFIT STRATEGY – SECOND NARROWS BRIDGE

The four-lane Second Narrows Bridge was constructed in the 1950s. The bridge consists of a steel truss girder that spans the main deck over the Burrard Inlet and a pre-stressed concrete deck above the approach piers, north of the inlet (See Figure 2). On the north side, the bridge is supported on shallow and piled foundations underlain by generally compact, mountain stream, deltaic soils deposited by the Seymour River. On the south side, the bridge is supported on foundations constructed on bedrock.

2 Figure 1. Key Plan

Figure 2. An Aerial View of Second Narrows Bridge (Looking East)

The Ministry of Transportation retained Golder Associates to provide geotechnical input to complete a Seismic Safety Retrofit for the Second Narrows Bridge. The work has been carried out in several phases since 1991. Some of the initial work undertaken involved a feasibility assessment of a parallel new bridge and upgrading of the existing bridge. Subsequent work considered upgrading the existing bridge only, and conducting a pilot densification program to assess feasibility of ground improvement. According to the Ministry’s guidelines, the Safety Retrofit has been carried out for an earthquake with a return period of 475 years. The bridge site was characterized by means of a uniform hazard, firm- ground response spectrum that corresponds to a PGA of 0.20 g and a design earthquake of magnitude M7.

3 4 FIELD INVESTIGATIONS

Golder conducted geotechnical and geophysical field investigations along the bridge and at specific bridge piers to establish the subsurface soil conditions underlying the foundations. These included conventional drilling with rotary methods, Becker Penetration percussion testing with energy measurements, downhole shear wave velocity testing, seismic refraction profiling, and Sonic drilling. Results of these field investigations were used to establish the distinct soil stratigraphic units and to derive engineering properties applicable to the different units. Becker percussion drilling was carried out since the site is underlain by river channel deposits that are generally coarse-grained. However, low permeability layers of soil within the generally coarse-grained overburden soils are also present. These low permeability layers were identified with the help of Sonic Drilling. This was considered an important issue given that, as has been demonstrated by laboratory tests conducted under controlled conditions, low permeability soil layers can lead to a post-liquefaction behaviour that is more severe than the commonly assumed “undrained” condition (Atigh & Byrne, 2000). The SPT N60 profiles established for the site are shown on Figure 3.

Figure 3. Profiles of SPT N60 Compiled for the Site

4 5 SEISMIC PERFORMANCE OF FOUNDATION SOILS

The liquefaction susceptibility of the foundation soils was initially evaluated using the results of 1-D SHAKE analyses and Seed’s liquefaction resistance chart. The results of these analyses indicated that liquefaction would likely extend to depths varying from 15 to 20 m below existing ground surface. In these initial analyses, the liquefaction-induced lateral ground movements were computed using an equivalent seismic coefficient derived from energy considerations. The results indicated that soil liquefaction would have a significant effect on foundation performance; e.g. foundations supporting a number of piers were assessed as having a high risk of liquefaction-induced punching and bearing capacity failure. In addition, the ground surrounding several piers was predicted to undergo free-field lateral movements in the order of 1 to 2 m towards the inlet. Consequently, extensive ground improvement measures were considered as mitigative measures. More detailed ground response analyses were undertaken subsequently to optimise the ground improvement measures required for the bridge foundations. Here, a displacement-based design was pursued and ground improvement measures were considered at selected critical locations along the bridge, so that the ground movements resulting from soil liquefaction were within structurally tolerable values. The detailed analyses were carried out using the 2-D finite difference code FLAC that incorporates a sequential liquefaction-triggering module developed by Beaty & Byrne (1999). The analysis was carried out in the time-domain, allowing for liquefaction to be triggered in individual soil zones during shaking, and taking into consideration the soil-foundation interaction effects. A schematic representation of the method followed in modeling sequential soil liquefaction is shown in Figure 4. Liquefaction triggering is evaluated by tracking the dynamic shear stress history of each material zone or element. The irregular shear stress history caused by the earthquake is interpreted as a succession of uniform half cycles. Each half cycle is transformed into an equivalent number of cycles that correspond to the penetration resistance of the zone, and liquefaction is triggered when the summation of equivalent uniform cycles is 15. With the onset of liquefaction, the soil is softened and the shear strength reduced to its residual value. The residual shear strength was considered to be stress path dependent, varying between about 0.4 and 0.1 times the initial vertical effective overburden stress depending on whether the major principal stress is vertical or at 45 degrees to the vertical, respectively. The limiting shear strain was assigned a value varying between 4 and 20% depending on whether the sand was in a dense or loose state, respectively.

a Ground Surface 1. Determine (N1)60 of the soil element. tcyc Compute the static shear stress in the element.

a) tstatic See Figure (a). N1 tstatic

b) 2. Establish (tcyc/s’vo) required to trigger liquefaction in 15 cycles as a function of (N1)60. ) vo ’ Compute t15 = (tcyc/s’vo) * s’vo. s / M=7.5 cyc See Figure (b). t ( NLiq»15 Figure 4: A Schematic Cyclic Stress Ratio Representation of the Liquefaction Evaluation (N ) 1 60 Procedure c) 3. Compute the cyclic shear stress time-history of xy t Half the element from the dynamic analysis for each Cycle half cycle. cyc t See Figure (c).

tstatic cyc t

4. Convert tcyc of each half cycle to (tcyc/t15), and Time enter Figure (d) to estimate the number of cycles,

NLiq, required to cause liquefaction. Multiply NLiq )

15 t t by 2 to estimate the number of half cycles of cyc /

cyc d)

t needed to cause liquefaction. Compute the relative Weighting Curve contribution of the current half cycle to liquefaction,

i.e. 1/(2NLiq). Compute the equivalent number of cycles based on Normalized 15 cycles to liquefaction; Neq= 15/(2NLiq).

Cyclic Stress Ratio, Log ( 5. Sum the effect of each half cycle, i.e. SNeq. # of Cycles to Cause Liquefaction, Log NLiq Liquefaction is triggered when SNeq³ 15. 5 Several 2-D soil profiles were analysed. These included soil profiles parallel to the bridge axis extending from the south end of the bridge to the north abutment. In addition, another soil profile was developed to analyse the performance of Pier P12 in detail, focusing specifically on the local soil and foundation conditions. The 2-D analyses, where liquefaction was triggered sequentially, predicted that the actual extent of soil liquefaction would be considerably less than what was predicted using the simplified 1-D SHAKE analyses. Although both methods predict similar depths to which soils are potentially liquefiable, the extent of predicted liquefaction is different because of the inherent assumptions in the two methods. For instance, the simplified method assumes that liquefaction occurs simultaneously through out the total depth of potentially liquefiable soils. In contrast, in the sequential method the onset of liquefaction in the weaker soil layers isolates the remaining soil layers from further shaking. Hence, the actual impact of liquefaction on ground displacements is less than if there is no isolation effect. The ground displacements predicted with the sequential method varied from about 30 mm to 300 mm. These displacements are about 2 to 4 times less than the displacements predicted from the simplified methods. The results of the ground deformation analyses supplemented with numerical simulations of bearing capacity failure were utilized to determine optimum locations for ground improvement. Typical results showing the deformed grid computed at the end of shaking at Pier P12 are shown on Figures 5a and 5b. Based on headroom considerations, ground improvement using seismic drains was considered at certain pier locations. The final layout of the ground improvement program selected for the bridge is shown on Figure 6.

6 SUMMARY

The seismic retrofit strategy adopted to enhance the seismic performance of the foundations of a major bridge structure on Trans Canada Highway in Vancouver, Canada, is presented. In this case, the project consultants adopted a displacement-based design where ground improvement was considered at selected critical locations along the bridge and at certain pier locations rather than at each pier. The latter approach was estimated to be considerably more expensive than the strategy adopted. The approach was feasible for a safety level retrofit where the primary focus is to minimise the risk of bridge collapse rather than its functionality following the design earthquake.

JOB TITLE : (*10^1)

FLAC (Version 3.40) 5.000

LEGEND

3-Nov- 0 4:01 3.000 step 3235423 4.685E+02

1.000 EX_ 7 Values 0.000E+00 Concrete 1.000E+00 Liquefaction not allowed 2.000E+00 No liquefaction -1.000 8.000E+00 Liquefaction Triggered 9.000E+00 Liquefaction Triggered Exaggerated Grid Distortion Magnification = 1.515E+01 Max Disp = 6.837E-01 -3.000

-5.000

-7.000

Golder Associates Ltd. Burnaby, Vancouver Canada 4.800 5.000 5.200 5.400 5.600 5.800 6.000 (*10^2) Figure 5a: Deformed Grid at the End of Shaking – Near Pier P12

6 JOB TITLE : (*10^1)

FLAC (Version 3.40)

0.750 LEGEND Displaced node

3-Nov- 0 4:01 Element step 3235423 0.250 5.234E+02

EX_ 7 Values Concrete -0.250 0.000E+00 Liquefaction not allowed 1.000E+00 2.000E+00 No liquefaction 8.000E+00 Liquefaction Triggered 9.000E+00 Liquefaction Triggered Displaced Grid -0.750 Exaggerated Grid Distortion Magnification = 6.252E+00 Max Disp = 4.184E-01

-1.250

-1.750

-2.250

Golder Associates Ltd. Burnaby, Vancouver Canada 5.250 5.300 5.350 5.400 5.450 5.500 5.550 (*10^2)

Figure 5b: Deformed Grid at the End of Shaking – Near Pier P12 (Zoomed)

Figure 6: Final Ground Improvement Program Adopted for the Bridge Foundations

7 The magnitude of liquefaction-induced ground movements and the resulting foundation displacements were critical in determining a suitable seismic retrofit strategy. The strategy discussed herein was based on extensive 1-D and 2-D ground response analyses that took into consideration the effects of onset of liquefaction as earthquake motions were propagated. Based on results from these analyses, it was concluded that the sequence in which liquefaction was triggered had a considerable impact on the permanent lateral and vertical ground movements that might occur for this specific site.

7 ACKNOWLEDGEMENTS

The authors are grateful to the BC Ministry of Transportation & Highways for granting permission to publish the technical information associated with the safety retrofit of the Second Narrows Bridge.

REFERENCES:

Atigh, E. and Byrne, P. M. (2000), “The effects of Drainage Conditions on Liquefaction response of Slopes and the Inference for Lifelines”, Proceedings of the 14th Annual Vancouver Geotechnical Society Symposium on Lifeline Geotechnical Engineering, May 2000. BC Ministry of Transportation (2000), Seismic Retrofit Design Criteria, BC Ministry of Transportation and Highways, May 2000. Beaty M & Byrne, P. (1999), “A Synthesized Approach for Modeling Liquefaction and Displacements”, Proceedings of the Conference on FLAC Numerical Modeling in Geomechanics, Balkema, Rotterdam. Byrne P. M. and Anderson D. L. (1987), Earthquake Design in Richmond, BC – Version II, Soil Mechanics Series No, 109, Department of Civil Engineering, UBC, Vancouver, Canada. Rogers G (1998), Earthquakes and Earthquake Hazard in the Vancouver Area, Geological Survey of Canada Bulletin No. 525.

8