New resilient bridge on liquefiable ground over the Ōpaoa River, Blenheim

H.A. Hendrickson, G. J. Saul & P. Brabhaharan

WSP, New Zealand.

ABSTRACT

A 190 m long, 2-lane bridge has been constructed over the Ōpaoa River on SH1, Blenheim. The bridge is in an area of high seismicity, and the site experienced liquefaction and lateral spreading in the 2016 Kaikōura earthquake. SH1 is a lifeline route and the bridge has been designed to importance level 3 to provide good resilience. The challenge was to achieve the expected resilience without adversely affecting the existing nearby heritage bridge structure which is to be preserved and used for walking and cycling.

The key project objectives, put forward by Waka Kotahi New Zealand Transport Agency, were to provide a resilient structure, balance risks, protect the heritage values of the nearby existing bridge and provide high levels of performance.

The design incorporates a zone of ground improvement by stone columns with a layer of high strength geotextile over top. Abutments are supported on driven h-piles and piers are founded on large diameter bored mono-piles.

This paper presents a summary of the foundation options considered and the basis for selecting the adopted option. Options considered include combinations of bored or driven deep piles, ground improvements at abutments in the form of stone columns of various construction methods, deep soil mixing, soil replacement and continuous flight auger piles. Vertical reinforced earth walls and spill through abutments were both considered for the bridge approaches. The superstructure and span lengths were carefully designed to optimise performance.

Key learnings and observations from the design and construction process are presented including verification of stone column ground improvement, design compromises made to improve safety in temporary works, verification of driven pile capacity and observations from construction of the pier foundations.

introduction

The pre-existing Opawa River Bridge (which is now a pedestrian and cycleway bridge) was located on State Highway 1, just north of Blenheim. A seismic assessment of the bridge indicated that it has poor seismic resilience during significant seismic events, narrow carriageway widths, poor geometric alignment and is vulnerable to scour during flood events. A replacement bridge has been designed and constructed to modern design standards and high levels of resilience.

This paper describes the design options considered, construction considerations and challenges.

A plan of the site and proposed bridge alignment is shown in Figure 1.

Figure 1: Site Plan

project challenges and objectives

The previous bridge is a bow string arch type reinforced concrete structure constructed between 1915 and 1917, making it a category 1 heritage structure as defined by Heritage New Zealand. It was retained in place as a pedestrian and cycling bridge. It is 170 m long and 5.49 m wide between kerbs which is significantly below the Austroads minimum requirement of 7 m for a two-lane bridge. Traffic modelling shows poor levels of service in both the morning and afternoon peaks, largely due to wide vehicles being unable to pass each other when meeting head on. Furthermore, wide vehicles are required to cross the centreline which poses safety risks to other road users and causes other traffic to have to slow down or stop.

The bridge is in an area of high seismicity, is of relatively heavy weight, and is vulnerable to damage in seismic and flood events.

The objective of the design is to provide a replacement bridge with high levels of seismic resilience, amenity and with geometrics and traffic safety designed to modern standards. The new bridge needs to address challenges associated with the poor ground conditions and high seismicity while balancing constraints associated with being adjacent to an existing heritage bridge, an alignment that runs through a campground, and the environmental constraints associated with working alongside and over a river.

Resilience

The 1915 arch bridge is located on State Highway 1 on the main national highway linking the North Island, through the Wellington to Picton Ferry, to Christchurch in the South Island. Therefore, the resilience of this bridge is of national significance. Locally, the bridge is immediately north of Blenheim town and if this bridge is closed, the alternate reroute access would be along State Highways 62 and 63 through Spring Creek, Rapaura, Renwick and Woodbourne (which may themselves be also closed due to poor resilience of bridges) and would add over 30 km and 30 minutes to any journey. Although there are smaller local roads with narrower bridge crossings, these are unlikely to be suitable for heavy vehicles or they themselves would be vulnerable to damage in an earthquake event.

The area is vulnerable to significant ground shaking and liquefaction due to the loose alluvial deposits present, and liquefaction and associated subsidence and lateral spreading is expected to affect the existing bridges pier and abutment foundations. A lack of restraint at the end bearings makes the existing bridge vulnerable to bridge span failure and movement of the heavy spans in the longitudinal direction causing shearing in the abutment piles.

The 2016 magnitude 7.8 Kaikōura earthquake caused only low levels of ground shaking of about 0.25g in Blenheim, extensive liquefaction was observed at the bridge site and lateral spreading was limited given the low ground shaking, see Figure 2. In larger earthquakes, with higher ground shaking, greater liquefaction and lateral spreading can be expected which is expected to cause damage to the existing bridge.

The flood resilience of the existing bridge is also very low with the centre pier of the bridge determined to be significantly comprised due to scour in a 1 in 100-year AEP flood.

An earthquake or flood event could lead to compromise of the structural integrity and closure of the bridge. Replacement with a new bridge would take a long time, leading to poor resilience.

Figure 2: Observed evidence of Liquefaction in the 2016 Kaikōura earthquake, with the old Opawa River bridge in the background

The new bridge was designed to Importance Level 3 (IL3) of NZS 1170.5, and to the requirements of the Bridge Manual 3rd edition (NZTA, 2016) which was updated based on learnings from the 2010-2011 Canterbury earthquake sequence (Wood et al, 2012). Early focus on resilience is important to achieve greater resilience for new infrastructure, and to avoid excessive cost (Brabhaharan, 2012), and therefore a key focus in developing design concepts was to select a bridge form and design concepts that would provide the level of resilience expected for the bridge.

DEVELOPMENT OF DESIGN CONCEPTS

Resilience was a key consideration in the development of design concepts and the selection of the bridge form. Lessons from damage to bridges in the Canterbury earthquakes in similar liquefaction conditions indicated that damage typically occurred to abutments and piers where they were positioned close to the river banks (Brabhaharan, 2012). Span arrangements were considered to minimise the numbers of piers close to river banks, and the design uses robust bored reinforced concrete piles for the piers, similar to those adopted for the Ferrymead bridge replacement after the earthquakes (Kirkcaldie et al, 2016). Ground improvement was adopted at the abutments to provide protection against liquefaction and reduce slope displacements. A piled abutment foundation was adopted to remove bridge loads from the embankment, further decreasing the potential for slope displacements.

The final structural form that was adopted includes 30 m long single hollow core bridge beams founded on piers supported by 2.1 m diameter mono piles, see Figure 3. The bridge abutments are founded on driven H-pile foundations (10 per abutment) within a zone of ground improvement comprising stone columns and an overlying layer of high strength geotextile and a drainage blanket to help dissipate excess pore water pressures. The pier piles near the river banks were designed to resist the full crust load of lateral spreading, as the soils surrounding them are not treated with ground improvement. Cyclic ground displacements were also considered in design.

The spans are simply supported at piers and abutments as this type of structure is more tolerant of potential vertical settlements in seismic cases compared to integral structural forms. The deck slab is continuous over the full length of the bridge with link slabs over the pier headstocks. The abutments are independent of the rest of the structure and the deck is free to translate in the longitudinal direction but restrained by shear keys in the transverse direction.

Figure 3: Elevation of Bridge

Shallow foundations for the piers were considered to offer low resilience as they would be vulnerable to scour, low bearing capacity in the surficial soils and unacceptably large settlements in seismic events where liquefaction occurs. Deep piled foundations can be designed to carry the large vertical loads and are resistant to many of the risks associated with shallow foundations. Due to the presence of cobbles and boulders in the intermediate gravel layers between 5 and 10 m depth (Unit 2) and the high levels of noise and vibration associated with driving the relatively large piles (or pile groups) that would be required to support the piers, large diameter bored piles were adopted at the bridge piers.

The pier piles are stiff structural elements compared to the abutments and therefore attract most of the longitudinal structural inertia loading. This leaves the H-piles and the abutments relatively lightly loaded and allows them to provide reinforcement to the approach embankments, should lateral spreading occur.

Close interaction between geotechnical and structural engineers was required throughout all stages of design to ensure consistency in the seismic modelling, that the bridge structural form selected suited the ground condition and that the structure provided a high level of resilience.

ground and groundwater conditions

Site investigations including boreholes, seismic shear wave velocity and standard cone penetrometer tests (CPTs) and laboratory testing were undertaken in two phases to characterise the site and support the various stages of design. Ground conditions at the site were found to include interbedded alluvial silts, sands and gravels underlain by very dense gravels of the Wairau Aquifer at 20 m depth. The silts were generally found to be non-plastic and in a soft to firm condition. The sands and gravels above 20 m depth were generally found to be very loose to medium dense with intermediate layers of very dense gravels.

Hydrostatic groundwater was encountered at or near river level across the site and slightly artesian groundwater was encountered in the Wairau Aquifer.

Potentially liquefiable soils include the non-plastic silts and loose and medium dense sands and gravels which are held in a sand matrix. Potentially liquefiable layers were identified at depths of up to 15 m below the water table level in Units 1 to 3 in Figure 4. The presence of liquefiable soils and the potential for lateral spreading at the site was confirmed during the 2016 Kaikōura earthquake where sand boils, ground subsidence and cracking were observed on site, see Figure 2.

Geotechnical assessment during concept design identified potential for flow liquefaction at the margins of the river, with the potential for lateral spread in two directions at the southern abutment and in one direction at the northern abutment. Therefore, lateral spread was a critical consideration in design.

An engineering geological section of the site is shown in Figure 4.

Figure 4: Geological Section

foundation concepts considered

Stability analysis of the new 6 m high approach fills and abutments with liquefied residual soil strengths in the underlying liquefiable soils confirmed that abutment instability can be expected in post liquefaction static cases and seismic vertical and lateral displacements which exceed the 50 mm limit set out in NZTA (2013). Mitigation was considered necessary to improve resilience of the structure and approach embankments in the vicinity of the abutments.

A reinforced soil embankment and piled abutment were both considered for restraining abutment movements. A reinforced soil embankment is flexible and tolerant of large vertical and horizontal movements but would require ground improvement to achieve bearing capacity requirements. Due to the skewed alignment, structural loads do not act along the alignment of the bridge making it impossible for the abutment piles alone to resist the large soil loads. Therefore, ground improvements were adopted.

Ground improvement options considered included; stone columns constructed by several different types of installation methods including soil replacement with stone or cement, deep soil mix columns and continuous flight auger piles in a grid or lattice layout. Due to the silty nature of the soils at the site, vibroflotation columns were considered unsuitable. Columns constructed using a steel casing which is then filled with stone and withdrawn as the aggregate is rammed into place or compacted with a displacement auger were considered most suited to the soil conditions. Stone columns also offered the advantage of lower costs and simpler construction when compared to other ground improvement methods considered.

The stone column methodology adopted was McMillan Civils displacement auger method. This method is vibration free.

Pile supported abutments with spill through embankments were adopted with a zone of stone column ground improvement at each abutment. The stone column target depths were optimised against potential slip surfaces and associated expected lateral displacements. Stone columns extend from approximately 1 m above water level to 10 m depth on the northern side and to 6 m depth on the southern side. They are overlain by 300 mm thick drainage blankets with two orthogonal layers of high strength geofabric to restrain lateral ground movements, see Figure 5. The combination of vertical piles and ground improvement at the abutments provided the resilience requirements for this bridge and operational continuity requirements of the Bridge Manual and any damage would be readily repairable following a design level earthquake. The solution provides a high level of resilience for relatively moderate costs and was accepted by the Client as a prudent investment.

Figure 5: Section showing ground improvement and piles and the abutments

design and construction

Temporary Works

In order to construct the high strength geotextile reinforcement at the required level, a temporary anchored sheet pile wall with a retained height of up to 4 m was required at both abutments. These walls supported the existing state highway which remained open throughout construction, and enabled the footprint and effectiveness of the zone of ground improvement to be maximised. The alignment of the sheet pile walls relative to the state highway had safety implications for the contractor and road users, with a tension between the design and construction related factors.

The position of the temporary walling was negotiated and optimised during construction. Minor reconfiguration of the stone column zone was required including, shortening of some columns due to clashes with the gravel hopper for the stone column rig and the top of the wall, and repositioning of some columns to satisfy the needs of the design and improve safety onsite.

Stone Column Ground Improvement

Design

Stone column ground improvement targeted improvement of the potentially liquefiable sands and silty sands of Unit 1 and the upper and liquefiable portion of unit 2 (refer to Figure 4).

Extensive discussions regarding different ground improvement options were held through an early contractor involvement (ECI) process to ensure the best outcome for the project. Based on the ground condition information available and conditions onsite, non-vibratory displacement columns were adopted. The final stone column configuration involved a zone of 800 mm equivalent estimated diameter columns positioned at 1.6 m centre to centre spacings on a staggered triangular grid layout. This resulted in an area replacement ratio of 29%.

The expected levels of ground improvement for the stone columns were determined based on the outcomes of Christchurch based improvement projects that WSP had previously designed. The key findings which were used are summarised in Table 1.

Table 1: Typical Ground Improvement Allowed for in Design

Our experience with this type of stone columns has been that post installation testing indicates significant mitigation of ground subsidence but can indicate some silty soil layers could remain potentially liquefiable. Allowance for some liquefaction between columns and associated lateral displacement was made in the design. The abutment piles are designed for the associated increased lateral demand taking into consideration the reinforcing effects of the H-piles on the embankment.

Ground Improvement Construction

Stone columns were constructed using a non-vibratory displacement auger method by specialist sub-Contractor, McMillan Civil Ltd. A site-specific stone column trial was specified, constructed and tested using CPT testing between columns and standard penetration testing down the columns, 35 days after installation. Initially, the trial results showed levels of improvement and densification did not meet design expectations in some layers. Possible reasons for this include the trial zone being too small to properly confine the soils and the increase in lateral confining stresses, may have been relieved when the testing was carried out too long after installation.

The trial was modified and repeated, and further verification testing was carried out, including cross hole shear wave velocity across columns within the production stone columns. This indicated measured improvements that aligned with design expectations. Verification CPT’s were positioned in the centre of the triangular grids, on grid lines and immediately beside columns. Cone tip resistance increased closer to columns, with some tests on grid lines and beside columns refusing on layers which were able to be penetrated in the centre of the grid. A summary of the pre and post installation CPT tip resistances and shear wave velocities are shown in Table 2.

The outcomes of the verification testing were compared to design expectations and the expected seismic performance of the bridge was checked and found to be acceptable.

Table 2: Ground Improvement Summary

Photographs of the stone column installation are shown below.

Photograph 1 – Displacement Auger
Photograph 2 – Installed columns showing heave around columns
Photograph 3: Completed columns showing sheet pile wall

Other Stone Column Construction Observations

Boreholes completed during the site investigations generally showed the ground on the northern side to consist of interbedded hard and soft layers, while on the southern side, the soft layers were much thicker and there were no hard layers within the depth range of stone column treatment.

Significant heave (up to 1 m³ per column) was observed on the northern abutment as stone columns were installed but not at the southern abutment. On the northern side, heave began as soon as the auger penetrated an intermediate hard layer and was observed to worsen as the auger penetrated and improved the hard layers. Heave is not ideal as it represents a loss of improvement effectiveness. It is expected that the heave on the northern side is due to the lateral confinement of the dense gravel layers (5 to 7 m depth) working against the outward displacement effort of the stone column auger and forcing soil vertically up the outside of the auger. The columns were constructed by first doing the perimeter columns then completing the remaining columns in a spiral pattern towards the middle of the zone to maximise the ground improvement potential. Heave was shown to increase towards the centre of the zone and is inferred to result from the increased horizontal stresses and significant ground improvement.

Abutment Piles

The bridge abutments are supported on 10 no. 30 m long 310- H-piles driven into the very dense gravels (Unit 4). Design pile bearing capacities were determined using Meyerhof’s method where the skin friction is a function of the SPT N value and an ultimate end bearing capacity of 10 MPa was assumed based on the piles being of low displacement type.

The required vertical ultimate pile capacity to be proven onsite was 1200 kN per pile and a geotechnical strength reduction factor of 0.64 was adopted in design. Pile lengths were generally governed by the large depths of liquefiable soils.

Pile vertical capacities were verified on site using pile dynamic analysis (PDA testing) during the final part of the installation drive and the Hiley formula and subsequent CAPWAP analysis on 20 % of the piles. PDA testing allowed for use of a higher geotechnical strength reduction factor for design and led to savings in pile length. It also reduced uncertainty around the end bearing capacity, which is especially important given the presence of liquefiable soils and associated potential for negative skin friction loads on the piles. The Hiley formula was then calibrated using the CAPWAP results to verify the capacity of the remaining piles.

Values of maximum compressive stress and temporary pile compression measured during PDA testing confirm driving conditions to be in the medium to hard range in accordance with ASG (2002).

The skin friction and end bearing capacities calculated in design and measured using pile dynamic analysis are presented in Table 3 along with capacities calculated using the Hiley formula in accordance with AGS (2002). Values are shown for piles which PDA analysis was undertaken only.

Table 3: Pile Capacity Summary

The ultimate capacities calculated using traditional and recommended Hiley coefficients in accordance with ASG (2002) give varying levels of conservatism compared to the CAPWAP analysis results. This is considered to be due to conservatism built into the Hiley formula coefficients, given that it is a simplified method of calculation and is therefore subject to a wide range of uncertainties.

It can also be seen from comparison of the Meyerhof design and CAPWAP analysis values presented in Table 3, that the Meyerhof method used in design over predicted the pile shaft capacity but was close to the end bearing capacity measured in the CAPWAP analysis. It also is possible the CAPWAP results reported in Table 3 are low as the shaft capacity often increases with time following driving and is higher in subsequent re-drives. A re-drive was not carried out on this project as all piles achieved the required capacity on initial installation and testing.

When typical pile strength reduction factors are considered, a strength reduction factor of 0.5 applied to the Meyerhof capacities and a factor of 0.64 applied to the CAPWAP outcomes, reduces the difference between the two.

Pier Foundations

The bridge piers are founded on 2.1 m diameter, permanently cased bored piles founded in the very dense gravels (Unit 4). The piles are designed to resist negative skin friction and have small settlements in seismic cases.

Pile casings were installed using a combination of vibratory driving then hammer driving. Variation in ground levels onsite meant casing lengths needed to be varied to keep pile tops above ground level during construction. For simplicity, all casings supplied to site were the same length and were driven until the required top of casing level was achieved. All of the casings except one were driven to below the required founding level and a 0.5 m to 1 m long soil plug was left in place inside the casing following excavation.

The only exception to this were the piles in the water channel where the artesian head was observed to be the highest. At the location of these piles, the contractor used a 1 m higher pile casing and this was sufficient to prevent upward flow of groundwater causing softening of the pile base until the pile was poured.

Pile founding depths were verified by inspection of soil samples collected from 2 m depth intervals as the excavation advanced and from the bottom of the pile. Drillers logs were also compared with borehole logs from site investigation to ensure layers matched and adequate penetration into the founding layer had been achieved. Pile cleanout was achieved by allowing the suspended soil in the fluid column to settle then using a cleaning bucket to remove the loose sediments from the base. This was repeated until the cleaning bucket did not pick up loose sediment. In addition to this, concrete was poured through a pressurised tremie, to scour and displace any remaining loose sediments from the base of the hole.

conclusions

The Ōpaoa River Bridge is in an area of high seismicity and is a geotechnically challenging site. Key challenges included soft and liquefiable ground, lateral spreading and physical space constraints.

The replacement bridge was designed to provide a high level of seismic performance and resilience. Close interaction between Structural and Geotechnical Engineers was essential throughout all phases of design and was critical in achieving successful resilience in an economical manner.

Geotechnical models and analysis used in design are theoretical and testing during construction is limited and constrained by the construction programme. Care needs to be exercised during design to understand these risks, allow appropriate tolerances and carry out monitoring and verification to confirm the design during construction. In this case the ground risks were managed by the adoption of a resilient and accommodating foundation solution which allowed the uncertainties in ground conditions and level of ground improvement to be effectively managed.

Key to achieving a successful bridge with resilience was a focus on the resilience from the early stages of the business case, concept, preliminary and detailed design as well as construction.

references

Auckland Structural Group (2002) Piling Specification – Revision G.

Brabhaharan, P (2014). Lessons from Liquefaction Damage to Bridges in Christchurch and Strategies for Future Design. NZ Society for Earthquake Engineering Conference, Auckland.

Kirkcaldie, DK, Brabhaharan, P, Cowan, M, Wang, C, Hayes, G and Greenfield, L (2013). Ferrymead Bridge – from widening and seismic strengthening to replacement, a casualty of the Christchurch Earthquake. NZSEE Annual Conf. Existing Risks New Realities. Wellington. April 2013.

New Zealand Transport Agency (2013) Bridge Manual, 3rd Edition (effective from May 2013) and Amendment 1 in Section 6 (effective from September 2014).

Standards New Zealand (2004). NZS 1170.5 Earthquake actions.

Standards Australia (2009). AS2159 Piling – Design and Installation.

Wood, JH, Chapman, HE and Brabhaharan, P (2012). Performance of Highway Structures during the Darfield and Christchurch Earthquakes of 4 September 2010 and 22 February 2011. Report prepared for the New Zealand Transport Agency. February 2012.