Ground Improvement Case Histories: Compaction, Grouting and Geosynthetics

Rujikiatkamjorn

Part One

Physical Modification Methods Including Grouting, Compaction, and Drainage

Chapter 1

Ground Improvement for Mitigating Liquefaction-Induced Geotechnical Hazards

Dharma Wijewickreme¹,²; Upul D. Atukorala²    ¹ Department of Civil Engineering, University of British Columbia, Vancouver, B.C., Canada

² Golder Associates Ltd., Vancouver, B.C., Canada

Abstract

Liquefaction of foundation soils imposes geotechnical hazards primarily in the form of loss of bearing capacity and permanent ground deformations. Foundations in liquefiable soils need to be designed to withstand these hazards, or ground improvement measures need to be implemented to mitigate the resulting impacts. Five engineering case histories are presented where ground improvement measures were undertaken to minimize the liquefaction-induced geotechnical hazards. The project sites corresponding to the case histories are located within or in the vicinity of the Greater Vancouver region of British Columbia, which is one of the zones of highest seismic risk in Canada. Details pertaining to site seismicity and subsurface conditions together with the design philosophy are presented to provide the necessary background information for the case histories. In addition, details of the construction equipment, postground improvement performance, and structural, geotechnical, and environmental monitoring that were undertaken are presented and discussed. The case histories correspond to ground improvement carried out using vibro-replacement, compaction grouting, and deep dynamic compaction methods. The protected foundations and/or structures belong to key industrial plants, highway systems, and energy networks.

Keywords

Soil Liquefaction

Ground Improvement

Deep Dynamic Compaction

Vibro-Replacement Stone Columns

Compaction Grouting

Gravel Compaction Piles

Acknowledgments

The authors are grateful to the B.C. Ministry of Transportation, Terasen Gas Utility Ltd., and Lafarge Canada Inc. for granting permission to publish the technical information associated with the case histories.

1.1 Introduction

Frequent occurrence of devastating seismic events around the world has resulted in a remarkable increase in the public interest toward earthquake preparedness. The known potential for disruption to structures and facilities has encouraged the owners to protect their assets from earthquake hazards. The seismic evaluation/upgrading programs undertaken over the last 20 years by lifeline owners in North America and Japan serves testimony to the significance of this subject (TCLEE, 1998; Wijewickreme et al., 2005). Experience from past seismic events indicates that earthquake-induced permanent ground displacements and/or loss of bearing capacity are some key geotechnical hazards to structures located at sites underlain by liquefiable soils (O’Rourke and Hamada, 1992; MCEER, 1999). After identification of the geotechnical hazards and the resulting vulnerability of a given structure, a combination of structural retrofitting and/or geotechnical remediation (ground improvement) is often considered in the design of mitigative measures.

Historically, ground improvement has been used as a means of improving the postconstruction bearing capacity and settlement performance of soils under static loading conditions, and a variety of ground improvement techniques have evolved in the past few decades (Mitchell, 1981; Japanese Geological Society, 1998). In addition to resisting static loads, some of the ground improvement measures have been effectively used to retrofit facilities that are located within, or that have foundations supported on, liquefiable soils. These measures include dynamic deep compaction, vibro-replacement using stone columns, compaction piling, explosive compaction, and compaction grouting.

The observed performance of sites following major earthquake events—for example, 1964 Niigata (Niigata, Japan), 1995 Hyogoken Nanbu (Kobe, Japan), 1999 Kocaeli (Turkey), 2001 Nisqually (Washington state, U.S.)—indicates that the sites with improved ground had generally less susceptibility to earthquake-induced ground deformations and resulting damage than the sites that had not been densified (Mitchell et al., 1998; Hausler and Sitar, 2001; Hausler and Koelling, 2004).

Amid these ground improvement efforts, including the examination of past earthquake damage and postearthquake operations, there is a need for more documentation of approaches and illustrative case histories related to the use of ground improvement. Clearly, advances in the state of practice in seismic evaluation and retrofit of facilities require dissemination, particularly within the structural and geotechnical engineering disciplines.

With this background, and drawing from a number of case histories from Greater Vancouver, British Columbia (B.C.), Canada, this chapter illustrates several key facets and considerations in the engineering of ground improvement to mitigate liquefaction-induced geotechnical hazards. The sites of the case histories are situated within one of the zones of highest seismic risk in Canada (NBCC, 1995). The region encompasses significant areas underlain by marine, deltaic, and alluvial soil deposits, some of which are considered to be susceptible to liquefaction and large ground movements when subjected to earthquake shaking. Seismic performance of the structures and lifelines located within such weak ground conditions has been of particular concern to the region at large.

The following aspects are specifically addressed:

• Current approaches for the evaluation of seismic vulnerability, including identification and prediction of geotechnical hazard (prediction of earthquake-induced ground deformations)

• Influence of soil conditions (including key controlling parameters and features), presence of site constraints, and environmental concerns in governing the selection of the most appropriate ground improvement method

• Ground improvement schemes/configurations used in addressing typical engineering situations

• Verification testing for quality control

• Monitoring of existing facilities during adjacent ground improvement construction

Examination of case histories involving observed field performance during past earthquakes is not included in the scope of this chapter.

Because all the case histories presented herein emanate from one general geographic area, some common background information related to the Greater Vancouver region is also included in the following sections as a part of the introduction. The approaches used in the assessment of earthquake-induced geotechnical hazard, as well as the philosophy adopted in the engineering design and evaluation of seismic retrofit measures, were found to be generally applicable to all the case histories. As such, this information has also been presented concisely in this introductory section.

1.1.1 Greater Vancouver region of British Columbia

More than 2 million people live in the Greater Vancouver region of British Columbia, Canada (see Fig. 1.1 for approximate location). The region covers a triangular-shaped area of about 3000 km² bounded by the Coast Mountains to the north, the Cascade Mountains to the south and southeast, and by the Strait of Georgia to the west. The Fraser River extends through the area and has developed a delta some 30 km long and 25 km wide.

Figure 1.1 Location plan.

Regional surficial geology

The regional surficial geology of the area has been mapped in detail by the Geological Survey of Canada (Armstrong, 1976, 1977; Armstrong and Hicock, 1976a,b). Glaciers repeatedly covered the area during the Pleistocene era, resulting in deposition of relatively competent glacial till and proglacial sands. Since the retreat of the last glaciers, more recent sediments (Holocene or postglacial deposits) associated with the Fraser River and other watercourses have been laid down. Channel fill and flood-plain deposits cover almost all of the low-lying areas within the Greater Vancouver region. These recent sediments are relatively unconsolidated and are judged to provide a medium for ground motion amplification and also considered susceptible to liquefaction when subjected to earthquake shaking.

Regional seismicity

The seismicity in the area results from the thrusting of the offshore Juan de Fuca plate beneath the continental North American plate. The subduction zone is located off the west coast of Vancouver Island. There are three distinct sources of earthquakes: (1) relatively shallow crustal earthquakes (depths in the order of 20 km); (2) deeper intraplate earthquakes (~ 60 km deep) within the subducted plate; and (3) very large interplate earthquakes, also referred to as megathrust or subduction earthquakes. Earthquakes within the first two categories (crustal and intraplate) have been recorded at regular intervals over the last several decades in the area. The largest recent earthquakes are those near Campbell River, B.C. in 1946 (M = 7.3), near Olympia, Washington in 1949 (M = 7.1), near Seattle/Tacoma, Washington in 1965 (M = 6.5), and near Nisqually, Washington in 2001 (M = 6.8).

Such earthquakes are commonly included in probabilistic and deterministic seismicity models. Large subduction earthquakes have not occurred in, or near, British Columbia in historic time. However, there is geological evidence to suggest that they have occurred in the past (possibly at 300–400 year intervals), and the measured accumulation of strain between the tectonic plates suggests that they should be expected in the future (Clague and Bobrowsky, 1994). Although the magnitude and duration of a large subduction earthquake is expected to be greater than the crustal and intraplate events, the intensity of ground shaking is not, due to the greater epicentral distance from the Greater Vancouver region.

The primary concern with respect to the subduction earthquake is the duration of shaking, expected to be in the order of 2 min, or more than four times that of the intraplate earthquakes. The National Building Code of Canada that was in effect at the time the project described later in this chapter were completed did not address the subduction earthquake scenario explicitly. However, owners of critical facilities often considered the impact of subduction loading on structures in addition to the seismic design provisions in the NBCC.

The seismic hazard for the area is evaluated primarily using existing probabilistic models of the region (e.g., Cornell, 1968; McGuire, 1976; Adams and Halchuk, 2004), while a deterministic approach is used to evaluate the effects of a large subduction event. The outputs from the probabilistic analyses typically consist of peak horizontal ground accelerations (PHGA) at firm-ground level, and magnitude and distance contributions corresponding to the hazard. Form ground PHGAs for the 1-in-475-year return period level, which was the return period embodied in the National Building Code of Canada that was in effect at the time the projects described later were undertaken, typically range from about 0.15-0.25 g for the region.

The predicted PHGAs at firm-ground level for a selected site in the area are presented in Fig. 1.2 to illustrate the variation of anticipated ground motions with the probability of occurrence. In some areas, firm ground is located at depths in the order of ~ 250 m below the ground surface, overlain by relatively soft, loose soils. The ground motions are expected to significantly amplify during transmission through these soft soil areas, and the estimated design PHGAs at ground surface levels of the order of 0.3 g are not uncommon.

Figure 1.2 Predicted ground motion at firm-ground level for a selected site in the region.

1.1.2 Assessment of earthquake-induced geotechnical hazard

The assessment of site-specific geotechnical hazards forms an important part in the assessment of overall seismic vulnerability of a given structure as well as in the development of potential mitigative measures to reduce the risk of damage to acceptable levels. As such, for a given site, the site-specific studies generally include the following key steps:

1. Geotechnical investigation to understand the site-specific soil and groundwater conditions

2. Assessment of site-specific ground motion parameters

3. Prediction of site-specific ground response for the identified seismic risk levels including assessment of site-liquefaction potential

4. Assessment of the geotechnical stability of the site

5. Assessment of earthquake geotechnical hazards using empirical and/or mechanistic approaches (e.g., seismic slope stability, liquefaction-induced ground movements, postearthquake bearing capacity)

In most instances, estimating earthquake-induced permanent lateral ground displacements becomes one of the critical steps in the site-specific seismic assessment. The methods available for the computation of earthquake-induced permanent lateral ground displacements can be broadly classified into (1) empirical approaches developed based on measured displacements (e.g., Bartlett and Youd, 1992; Power et al., 1998; Youd et al., 1999; Rauch and Martin, 2000; Bardet et al., 2002; Youd, 2002) and (2) mechanistic approaches that rely more on the principles of engineering mechanics (Prevost, 1981; Finn et al., 1986; Byrne et al., 1992, 2004).

As noted by Glaser (1994) and Youd (2002), the estimation of earthquake-induced ground deformations, particularly from the viewpoint of regional assessments, still relies heavily on empirical correlations. Even with the advantage of being based on an actual database of ground displacements while accounting for many physical parameters governing the free-field ground displacements, empirical methods do not incorporate the deformation modulus and shear strength of liquefied soil in the computations. These deformation characteristics are considered key factors influencing the magnitude of ground displacements. Empirical methods are also unable to estimate ground displacement patterns on the surface and at different depths and, therefore, they cannot account for the presence of human-made site features (i.e., zones of ground improvement).

These parameters are often a necessity in the assessment of seismic vulnerability and design of mitigative measures. Numerical approaches based on soil mechanics principles provide a means to estimate displacement patterns but suffer from lack of rigorous calibration with actual earthquake data. Based on these considerations, in the case studies described herein, the assessment of ground displacement geotechnical hazard was typically undertaken using a hybrid approach combining both numerical and empirical methods (Wijewickreme et al., 1998). In this, the estimated free-field ground displacement hazard using empirical approaches was used to calibrate the results from site-specific analyses of the site derived using numerical/mechanistic approaches.

1.1.3 General philosophy of seismic retrofit

In general, there are only four options to improve the seismic performance against an identified geotechnical hazard: (1) avoid the hazard by relocation, (2) isolate the structure from the hazard, (3) accommodate the hazard by strengthening the structure, and (4) reduce the hazard using ground improvement.

Typically, all of the preceding options are considered in developing retrofit concepts. When ground improvement is considered as the desired option, the selection of the most suitable remedial option is governed by many factors including, but not limited to: soil conditions, equipment/space restrictions, issues related to the protection of existing structures during ground improvement, operational constraints, environmental regulatory requirements, and land availability.

1.2 Case history 1: ground improvement using vibro-replacement at a site with buried gas pipelines

This section presents a case history on the seismic upgrading of a buried natural gas pipeline gate station site in Vancouver, B.C. The gate station is part of a major natural gas transmission system. Given the lack of redundancy in certain areas of the system, a very low risk of disruption to the gas supply is considered acceptable to the owner. Prevention of loss of pipeline pressure integrity under earthquake loading corresponding to an annual probability less than 0.05% (equivalent return period 2000 years) was used as the key seismic performance criterion in judging the acceptability of pipeline performance. Based on a regional study, the gate station was assessed to be highly vulnerable to damage under seismic loading. Liquefaction-induced ground deformations were identified as significant hazards to the pipelines entering the gate station and associated facilities. The design and construction of the seismic upgrading work at the site was undertaken between 1995 and 1997.

1.2.1 Site description and subsurface soil conditions

The gate station compound is generally rectangular in plan (~ 100 m × 75 m) and located on the north bank of the north arm of the Fraser River (see Fig. 1.3). As illustrated, two transmission pipelines (NPS 20 and NPS 24)¹ enter the gate station below the riverbed from the south.

Figure 1.3 Site plan showing existing structures, pipeline configurations, and geotechnical testhole locations.

The site topography within the station compound and also in the east–west direction is generally flat. Prior to ground improvement, the riverbank sloped down toward the south at slopes ranging from 1H:1V–3H:1V (horizontal: vertical) within the riprap area, which extended to about 6 m below crest level. The riverbed below this level sloped southward at an average gradient of about 8% to the horizontal.

Figure 1.4 presents a profile illustrating the inferred soil stratigraphy at the gate station site, developed based on a geotechnical field investigation comprising both onshore and over-water geotechnical drilling. A combination of the methods of electric cone penetration testing (CPT), mud-rotary drilling, and solid-stem auger drilling was used in the field investigation. The upper soils within the station consist of about 2–3 m of loose to compact sand to sandy silt fill material. The upper fill materials in the northern part of the gate station are underlain by a layer of very soft to soft silt (liquid limit, WL = 38%; plasticity index, IP = 11%; water content, w = 40%) extending to depths of the order of 6–8 m below the ground surface. The silt zone is underlain by a compact to dense sand stratum, which, in turn, was found to overlie a stratum of very dense sand and gravel at a depth of about 9 m below the ground surface.

Figure 1.4 Profile of soil stratigraphy, predicted zone of potential liquefaction, and alignment of 610 mm-diameter pipeline.

Within the southern shoreline of the gate station, the soils underlying the upper fill materials primarily consisted of loose to compact sand extending to depths of up to 12 m below the ground surface. Underlying these soils, compact to dense sand with some gravel was encountered. These strata are underlain by dense glacial till-like material that was encountered at a depth of about 14 m below the ground surface. CPT testing within the river adjacent to the site also indicates the presence of sandy soils, below a 2 m thickness of silt and clayey silt, and extending down to a depth of about 9 m below the riverbed. These materials are underlain by a compact to dense soil stratum. The groundwater level at the site was assessed to be located at depths of about 1–3 m below the ground surface at the site.

1.2.2 Geotechnical performance under earthquake loading

The seismic response of the site was analyzed using the one-dimensional wave propagation program SHAKE (Schnabel, 1972). Charts developed by Seed et al. (1985) were used to assess the liquefaction potential of the site soils. The results indicated that the loose to compact sands in the southern portion of the site would liquefy under the levels of seismic loading investigated. For the seismic risk levels corresponding to 1:100-, 1:475-, and 1:2000-year return periods (with corresponding peak horizontal firm-ground accelerations 0.09 g, 0.20 g, and 0.34 g, respectively), the loose sandy soils at the site extending to a depth of ~ 12 m was found to be potentially liquefiable (see Fig. 1.4). An earthquake magnitude of M7 was used in the liquefaction assessment based on the anticipated magnitude contributions corresponding to the considered seismic hazard level. Based on the Chinese criteria (Marcuson et al., 1990) for the assessment of liquefaction susceptibility of fine-grained soils, the risk of liquefaction of the silty strata located within the northern portion of the site was classified as low.

The postliquefaction stability of the gate station compound was analyzed, using both circular and noncircular failure surfaces, to investigate the potential for a flow slide condition at the site. The postliquefaction shear strength parameters for potentially liquefiable zones were mainly selected based on laboratory postcyclic monotonic simple shear test data reported by Pillai and Stewart (1994) and Sivathayalan (1991). Based on this information, postliquefaction shear strength equivalent to 20% of the initial effective vertical stress was assumed for the stability analyses. Potential slip zones with significant encroachment into the station compound (failure zones extending landward about 30 m from the riverbank) were computed to have a postliquefaction factor of safety less than 1.0, even without application of any seismic inertia forces. This suggested a high risk of a flow slide, leading to very large deformations for the southern part of the site, for earthquake loadings, corresponding to all three risk levels.

Ground displacement analyses were conducted using a number of methods available at the time to assess the magnitude and patterns of the relative ground movements in the area north of the predicted flow slide zone. In particular, the liquefaction-induced free-field lateral ground displacements were calculated using the computer program developed by Houston et al. (1987), sliding block method by Newmark (1965), the empirical MLR method developed by Bartlett and Youd (1992), and a mechanistic finite element approach by Byrne et al. (1992).

The predictions from all analysis techniques indicated that, for the seismic loadings corresponding to all the risk levels considered in the study, large lateral ground displacements (in excess of 3 m) toward the river would influence an area extending to about 30 m north from the crest of the riverbank. The ground displacements for the nonliquefiable silty zones within the northern half of the site were computed to be less than about 0.1 m. Along with lateral ground movements, significant vertical ground movements were expected to occur within the southern area of the site due to translation movements of the soil mass.

1.2.3 Seismic performance of gate station piping

The vulnerability of the piping at the site was assessed using the results of a previous regional structural vulnerability analysis of the pipeline system at the site. This structural vulnerability analysis had been carried out using the computer code ANSYS (product of Swanson Analysis Systems, Inc.), in which the soil is modeled as a series of bilinear springs. The approach is discussed in more detail by the American Society of Civil Engineers (1984), the ALA (2001), and Wijewickreme et al. (2005). The outcome of this previous analysis was supplemented with site-specific soil strength information, knowledge gathered from past experience with detailed evaluations of similar configurations, and observation of pipeline performance in past earthquakes. Based on the soil type and soil strength information, approximate pipeline capacities in terms of ground deformations for the assessment of the gate station piping were computed, as summarized in Table 1.1. Although these estimates were considered approximate, they served as a means to judge the relative severity of the site-specific estimates of ground deformation on the pipelines.

Table 1.1

Summary of structural pipeline capacities in terms of acceptable ground movements

Maximum computed ground deformations derived from the geotechnical analyses were compared with the computed structural pipeline deformation capacities. The computed large ground displacements and resulting differential displacements at the gate station from earthquake-induced liquefaction were found to exceed the estimated capacity of the pipelines by an order of magnitude. This indicated that the risk of damage to the station piping under earthquake loading was well above the owner’s acceptance criteria.

1.2.4 Evaluation of remedial options

Given the gross exceedance of the available pipeline capacity, the only remedial measures deemed practical for the gate station involved improving the ground conditions. Provided that ground improvements could reduce the potential earthquake-induced permanent ground deformations to less than about 15 cm, no modification of the existing station piping was judged to be necessary.

The effectiveness of ground improvement in reducing the liquefaction-induced ground displacements at the site was assessed again using slope stability and finite element analysis. A design concept assuming the densification of an inground zone to minimize the anticipated ground movements was investigated. The width of the densification zone was assumed to be 10 and 20 m and the treatment was assumed to extend to the predicted full depth of liquefaction. The results indicated that the introduction of an inground densified barrier, in the order of 15–20 m wide, would reduce the expected earthquake-induced ground movements below the structural deformation capacity of the pipelines. In addition to the ground improvement, the shoreline slope was configured to a gentler slope to improve the riverbank slope stability.

1.2.5 Ground improvement using vibro-replacement

Several methods were considered to improve the liquefaction resistance of the soils. The selection of the most suitable ground improvement technique was governed by several factors such as soil conditions, equipment space restrictions, pipeline protection issues, environmental regulatory requirements, and land availability. Based on an evaluation of these considerations, the method of vibro-replacement was considered to be the most suitable technique of ground densification for use at the gate station site.

In total, 273 stone columns were installed (using the method of vibro-replacement) in a triangular pattern at 3 m center-to-center spacing to cover the plan area shown earlier in Fig. 1.3 to improve the overburden soils. A poker-type V23 vibrator with a rated energy of 165 hp was used to install the stone columns using the top-feed method. All stone columns extended to the top of the underlying hard stratum to depths between 8 and 16 m below the existing ground surface, with an average depth of about 14 m. The average amperage output during construction of individual stone columns was about 150 A, with peak outputs ranging from 170–260 A.

Six stone columns, from the initially proposed densification pattern of 294 columns, had to be deleted due to concern that installation at these locations may result in unacceptable deformations of adjacent gas pipelines. Boulders, concrete, and timber obstructions were encountered during column installation at some locations, generally at depths of some 3–6 m below the existing ground surface preventing the installation of another 22 stone columns. Of these 22 locations, 7 columns were successfully installed at alternate locations by relocating within 1.5 m of the design location. In general, attempts were made to relocate stone columns rather than locally excavating the obstruction. This approach was adopted due to concern that some of the timbers encountered could extend within the gas transmission pipeline corridor, and disturbance of these obstructions would present unacceptable risk of damage to the buried pipes.

1.2.6 Postimprovement verification testing

Field verification testing was performed at selected centroids of the stone column pattern using the method of electric cone penetration testing (CPT) during the progress of the densification program. The results of the postdensification testing together with review of the stone column installation details indicated that the cone tip resistance (qt) values generally exceeded prespecified performance qt criteria (ranged between 100 bar and 125 bar for clean sand zones in the zone of potential liquefaction). Some of the initial CPTs, carried out within about 14 days from the time of stone column installation, indicated that the specified qt requirement was not satisfied in certain zones of silty fine sand (N.B.: qt requirements were corrected for silt content); however, repeat testing carried out in the same area after several weeks from the installation of the stone columns indicated that the qt values had increased significantly from the initial postdensification values, and met the specified criteria for silty sands (see Fig. 1.5).

Figure 1.5 Results of postdensification electric cone penetration tests at the centroid of adjacent stone column triangular patterns (comparison between results from tests conducted 13 and 34 days after installation).

1.2.7 Monitoring of facilities during construction

The owners’ operational requirements necessitated that the gas supply be maintained throughout the construction work at the site. Although there was some flexibility to reduce the gas pressure in the transmission pipeline that was closest to the vibro-replacement work area (and carry out the vibro-replacement work in a sequential manner), a complete shutdown of the pipelines was not feasible. To minimize the risk of pipeline distress due to the construction activities, the zone of vibro-replacement was kept at least 2 m away from the existing facilities. These included existing high-pressure gas transmission pipelines, pile-supported heaters/filters, and regulator building, as well as piping used for monitoring and supplying fuel gas. Field instrumentation, including slope inclinometers, deep settlement gauges, and piezometers, were installed prior to construction work to monitor the horizontal and vertical ground movements and groundwater level variations in the vicinity of the buried pipelines during the process of ground treatment.

The lateral ground movements were evaluated from periodic slope indicator probe measurements at four locations. The results indicated lateral ground movements of less than 25 mm at depths greater than about 6 m below the ground surface. Lateral ground movements between about 25 mm and 60 mm were noted within the upper 6 m. As expected, the predominant lateral movement of these upper soils was in a direction away from the zone of densification.

In general, ground settlements of < 25 mm were observed during installation of stone columns at lateral distances of > 6 m from the settlement gauge locations. Observed settlements of monitoring points within 5 m depth below the ground surface increased to between 70 mm and 230 mm during installation of stone columns located between 0.5 m and 1.5 m horizontally from the settlement gauge. The observed settlements remained less than about 70 mm for monitoring points at depths greater than 5 m below ground surface.

During the vibro-replacement work, field measurements were periodically carried out to measure the groundwater level in the piezometers to assess any potential buildup of pore water pressures in the soil mass near the riverbank. The measurements indicated that there was no significant buildup of pore water pressures due to vibro-replacement. The subsurface survey monitoring points on the gas transmission pipes were established by excavating down to expose the top of the pipe, installing a vertical flexible plastic pipe that extended above ground surface, and then backfilling around the gas line and plastic access pipe.

The survey crew was onsite during installation of stone columns within 6 m of sensitive facilities, and monitored a selected group of survey points in the immediate vicinity of the construction activity every 15–30 min. The site engineer was informed immediately if deformations in excess of 3 mm difference from preconstruction measurements were observed. The observed vertical movements are summarized in Table 1.2. The allowable movement tolerance limits to meet the operating requirements, shown in the table, were determined based on structural evaluation prior to vibrodensification, and the observed vertical movements during construction were all within the defined tolerance limits.

Table 1.2

Summary of survey monitoring of the gate station facilities

1.2.8 Environmental monitoring

As part of the overall construction monitoring, an Environmental Protection Plan (EPP) was incorporated in the vibrodensification and shoreline restoration contract. The EPP identified the nature and magnitude of potential impacts as well as associated risks to the environment and defined the minimum care, procedures, and contingency measures to be exercised by the contractor(s) for the protection of the environment during the construction period. Compliance with the regulations and conditions was facilitated by the preparation and distribution of an environmental emergency response card. An environmental monitoring program was implemented to monitor the contractor’s compliance with the requirements of the EPP.

1.3 Case history 2: ground improvement using stone columns in coarse-grained soils at a highway bridge crossing

This case history presents the foundation retrofit strategy adopted to enhance the seismic performance of the foundation soils of a major bridge on the Trans-Canada Highway in Vancouver, B.C. This 1.3-km-long, six-lane bridge was constructed in the 1950s when earthquake design considerations were less stringent in comparison to the current bridge design codes, and also at a time when phenomena such as soil liquefaction and lateral spreading were not well understood.

The primary focus was to minimize the risk of bridge collapse rather than maintaining functionality following the design earthquake. This specific retrofit strategy is categorized by the owner as a safety-level retrofit, and the design earthquake corresponds to a seismic event having an annual probability of exceedance of 1/475 (or a return period of 475 years). The site-specific ground motions were characterized by a uniform hazard firm-ground response spectrum with a peak horizontal ground acceleration (PHGA) of 0.20 g and a design earthquake of magnitude M7 representing 10–15 cycles of loading.

To achieve the owner’s project requirements, a displacement-based design approach was adopted. The approach involved assessing the anticipated seismic displacements of each bridge pier, and, in turn, undertaking ground improvement at locations that are identified as critical. The design phase was completed in May 2001, and the ground improvement measures were implemented in 2002.

1.3.1 Site description and subsurface soil conditions

The bridge spans across Burrard Inlet and is located about 10 km east of Vancouver city center. The north side of the bridge is underlain by coarse granular soils (i.e., primarily sand and gravel) that form the alluvial fan of the Seymour River, which discharges into Burrard Inlet (Fig. 1.6). Due to the coarse nature of site soils, a number of different techniques were used to characterize the subsurface soils. 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.

Figure 1.6 Bridge site—the alluvial fan can be seen on the left (i.e., north abutment).

Becker penetration testing (BPT) consists of driving a closed-toe steel pipe that is 169 mm (6.65 in) in diameter using an ICE 180 diesel hammer that delivers 11 kN m of rated maximum energy per blow. The number of blows required to drive the steel pipe were recorded over penetration increments of 0.3 m and these blow counts along with hammer energy levels were later converted into equivalent standard penetration test (SPT) blow counts following the procedures outlined by Harder and Seed (1986). The conventional SPT sampler, due to its relatively smaller diameter (50 mm, 2 in, outer diameter), could give rise to unrealistically high penetration resistance values in coarse granular soils. Therefore, BPT with a relatively larger diameter penetrating tool is often adopted in the characterization of coarse-grained soils.

The presence of low-permeability layers of soil within the generally coarse-grained overburden soils was considered a concern, and this aspect was investigated using the method of sonic drilling (www.prosoniccorp.com), where continuous samples of soils were obtained that provided information on soil stratification. This was considered important because low permeability soil layers can lead to a postliquefaction behavior that is more severe than the commonly assumed undrained condition (Atigh and Byrne, 2000). The predensification SPT N60 profiles established for the site are shown in Fig. 1.7. Some photographs of soil samples obtained from sonic testing carried out in the potentially liquefiable soil stratum are shown in Fig. 1.8.

Figure 1.7 Preimprovement penetration resistance profiles.

Figure 1.8 Photographs of continuous samples from sonic drilling (typical).

Based on ground response analyses, the foundation soils were identified as having a high risk of liquefaction under the design earthquake ground shaking. The analytical results indicated that soil liquefaction may extend to depths varying from 15–20 m below existing ground surface. For most of the critical foundations, liquefaction was not predicted below the pile foundations, and the main geotechnical hazard was noted to arise from liquefaction-induced lateral spreading. A ground improvement program involving the construction of inground densified zones (barriers) at selected critical locations along the bridge alignment was considered the most appropriate retrofit measure to mitigate this hazard.

1.3.2 Evaluation of the preferred mitigation option—pilot ground improvement program

Further to an assessment of the different feasible ground improvement methods and associated costs, ground improvement using vibro-replacement was selected as the most cost-effective method of treatment in areas where the available headroom permitted the use of vibro equipment. The proximity of the work to existing settlement sensitive bridge foundations, depth of treatment required, available headroom beneath bridge deck, and cost of treatment were critical factors in making this decision. Where headroom was a concern, it was decided to enhance drainage to permit rapid dissipation of the excess pore water pressures generated by the earthquake by installing drains or densifying ground using gravel compaction piles.

A pilot densification program was undertaken to densify a selected 10 m × 10 m test area using vibro equipment. The test area was located between the approach piers P9 and P10 some 6–7 m away from the outer edge of the foundations. Pier P9 measures 12 m × 20 m in plan and pier P10 measures 9 m × 26 m in plan. Both piers have massive foundations and a depth of embedment of about 3 m.

The pilot densification project was initially undertaken using an electric V-type vibrator (Wightman, 1991) that has a rated horsepower of about 165 hp, and utilizing the top feed method. Significant amperage build-up was noted during penetration of the upper 9–11 m of soils that comprised gravel, cobbles, and boulders, resulting in probe withdrawal prior to reaching the design depth of 18 m. The ground improvement contractor did not allow the V-type vibrators to reach amperage values much in excess of 200 A for more than a few seconds due to concerns of damaging the motor due to overheating. This led to significant problems as it caused the coarser particles to descend down the hole (as opposed to being washed out or pushed into the soil formation laterally) and form a nested layer in the midway depth of the intended column. Pullback actions of the vibrator further promoted the tendency for the coarse particles to fall into the hole. The end result was a zone of near impenetrable nested coarse particles that did not allow probe penetration without very high buildup of amperage and obstructed the densification of any loose soil layers located at depth.

Following discussions with specialty contractors and the project team, a decision was made to discontinue the use of the V-type probe and, instead, utilize a proprietary S-type vibrator (www.haywardbaker.com) that is capable of sustaining amperage readings as high as 300 A for longer durations without motor burnout concerns. The S-type vibrator was successful in penetrating the coarse-grained soils to the design depth at a vibrocolumn full-completion degree of 90% (in comparison to 70% with the V-type vibrator).

Postimprovement penetration resistance derived from BPT testing was well in excess of the specified (N1)60 = 28 blows/0.3 m. In essence, the pilot densification project confirmed that the required penetration resistance values can be achieved with the vibrocompaction method, provided that robust equipment that can sustain amperage readings as high as 300 A for longer durations in the order of 10–30 s are used.

1.3.3 Detailed engineering analyses

Following the pilot densification project, rigorous ground response analyses were undertaken to assess the effectiveness of inground densification barriers located at strategic locations along the bridge axes. The objective was to optimize the ground densification requirements considering the displacements that can be tolerated by the bridge foundations. Differential lateral displacements between adjacent footings in the order of 400–700 mm were considered acceptable from structural considerations.

Free-field ground response analyses as well as soil–structure interaction analyses were undertaken to assess the bridge foundation performance under the 475-year ground motions using the computer code Fast Language Analysis of Continua (FLAC, Version 3.4, 1998). A cyclic stress–strain module that was capable of simulating sequential liquefaction in accord with the commonly used liquefaction resistance chart (Seed et al., 1985; Youd et al., 2001) was used in the analysis of ground deformations. The results were compared with empirical methods of estimating ground deformations and were in good agreement. Details are presented elsewhere (Atukorala and Puebla, 2003).

For most of the critical bridge piers, liquefaction was not predicted below the pile foundations, and relatively larger pier movements under seismic loading were considered structurally tolerable. As such, it was possible to design an optimum ground improvement scheme where constructing inground barriers at selected critical locations was assessed to be sufficient to reduce the lateral displacements to acceptable levels. The alternative approach would have been to carry out ground improvement at each and every pier location, with significantly high installation costs.

1.3.4 Final ground improvement program

Ground improvement measures implemented for the bridge in 2002 included installation of vibro-replacement stone columns, gravel compaction piles, and seismic drains. Stone columns were used at piers P14 and P15, located in the over-water area of Burrard Inlet. Gravel compaction piles were used in the onland area between piers P11 and P12, where the upper 6 m of the site is underlain by coarse granular fill materials comprising sand and gravel with cobbles and possibly boulders. The seismic drains were installed at the onland pier P11 where the available headroom was limited for the use of vibro equipment. The depth of treatment varied from 10–15 m below existing ground surface or seabed.

Stone column installation (between piers P14 and P15)

Stone columns were constructed using the bottom-feed, wet method of construction. An S-type vibrator with a 165 hp energy rating mounted on a 60-ton crane was utilized for the installation of stone columns. The amperage buildup and depth of probe penetration were monitored during stone column construction. The contractor elected to install the stone columns using a square grid pattern at a center-to-center spacing of 3 m. A total of 365 stone columns were installed. Photographs taken during stone column installation are shown in Fig. 1.9.

Figure 1.9 Installation of stone columns using bottom-feed system ( left ) and loading of backfill stone to hopper ( right ).

Gravel compaction piles (between piers P11 and P12)

The gravel compaction piles were installed using a rig typically used for the installation of expanded-base piles. In this installation, at a given pile location, a steel drive tube was initially installed using a drop hammer capable of delivering a rated energy of 215 kN m (160,000 ft lb) per blow. A gravel shaft was then constructed using 0.07 m³ (2.5 ft³) of gravel compacted with a hammer capable of delivering a rated energy of 135 kN m (100,000 ft lb) per blow. The initial expelling of the gravel plug was carried out at a higher energy level of 215 kN m (160,000 ft lb) per blow.

The contractor elected to use an equilateral triangular pattern of gravel compaction piles with a horizontal center-to-center spacing of 1.95 m (6.5 ft) between adjacent piles. The initial target was to install all compaction piles using the 508 mm (20 in) steel tube. However, due to heavy penetration resistance encountered at a number of locations, a combination of 406 mm (16 in) and 508 mm (20 in) steel drive tubes were used. In total, 50 gravel compaction piles were installed using the 508 mm tube, and 96 piles were installed using the 406 mm tube.

The total length of steel casing used for the construction of gravel compaction piles was 18.3 m (60 ft). This length was 1.8 m short of the target depth of treatment. The use of the shorter casing length was accepted in view of the extent of compaction anticipated below the bottom of the casing due to expelling of the plug to an estimated depth of 5–6 diameters. Photographs taken during gravel compaction pile installation and stone backfill used are shown in Fig. 1.10.

Figure 1.10 Equipment used for installation of gravel compaction piles ( left ) and stone used in gravel compaction piles ( right ).

1.3.5 Postimprovement verification testing

The postimprovement testing was carried out using the Becker hammer. An HAV-180 rig with 169 mm (6.6 in) casing was used for all postimprovement verification testing. Bounce chamber pressure was measured with depth of probe penetration to obtain necessary parameters for the inference of equivalent SPT (N1)60 values. All of the measurements were recorded by the prime contractor and verified by the owner’s staff